diff --git a/.gitignore b/.gitignore
index 646de6fd..7734e2a4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -8,3 +8,4 @@ ext/.depend.mk
 *.pyc
 *~
 webdata/
+.cache/
diff --git a/gentle/__init__.py b/gentle/__init__.py
index 77a74e05..cfe17060 100644
--- a/gentle/__init__.py
+++ b/gentle/__init__.py
@@ -3,3 +3,4 @@
 from forced_aligner import ForcedAligner
 from full_transcriber import FullTranscriber
 from resample import resample, resampled
+from transcription import Transcription
diff --git a/gentle/diff_align.py b/gentle/diff_align.py
index a40e1269..70cc6dd0 100644
--- a/gentle/diff_align.py
+++ b/gentle/diff_align.py
@@ -3,10 +3,11 @@
 import os
 import sys
 
-import metasentence
-import language_model
-import standard_kaldi
-from resources import Resources
+from gentle import metasentence
+from gentle import language_model
+from gentle import standard_kaldi
+from gentle import transcription
+from gentle.resources import Resources
 
 
 # TODO(maxhawkins): try using the (apparently-superior) time-mediated dynamic
@@ -23,7 +24,7 @@ def align(alignment, ms, **kwargs):
     disfluency = kwargs['disfluency'] if 'disfluency' in kwargs else False
     disfluencies = kwargs['disfluencies'] if 'disfluencies' in kwargs else []
 
-    hypothesis = [X["word"] for X in alignment]
+    hypothesis = [X.word for X in alignment]
     reference = ms.get_kaldi_sequence()
 
     display_seq = ms.get_display_sequence()
@@ -36,17 +37,16 @@ def align(alignment, ms, **kwargs):
             word = hypothesis[a]
             if disfluency and word in disfluencies:
                 hyp_token = alignment[a]
-                phones = hyp_token.get("phones", [])
-                start = hyp_token["start"]
-                end = hyp_token["start"] + hyp_token["duration"]
-
-                out.append({
-                    "case": "not-found-in-transcript",
-                    "phones": phones,
-                    "start": start,
-                    "end": end,
-                    "word": word
-                })
+                phones = hyp_token.phones or []
+                start = hyp_token.start
+                end = hyp_token.start + hyp_token.duration
+
+                out.append(transcription.Word(
+                    case="not-found-in-transcript",
+                    phones=phones,
+                    start=start,
+                    end=end,
+                    word=word))
             continue
 
         display_word = display_seq[b]
@@ -55,28 +55,26 @@ def align(alignment, ms, **kwargs):
         if op == 'equal':
             hyp_word = hypothesis[a]
             hyp_token = alignment[a]
-            phones = hyp_token.get("phones", [])
-            start = hyp_token["start"]
-            end = hyp_token["start"] + hyp_token["duration"]
-
-            out.append({
-                "case": "success",
-                "startOffset": start_offset,
-                "endOffset": end_offset,
-                "word": display_word,
-                "alignedWord": hyp_word,
-                "phones": phones,
-                "start": start,
-                "end": end,
-            })
+            phones = hyp_token.phones or []
+            start = hyp_token.start
+            end = hyp_token.start + hyp_token.duration
+
+            out.append(transcription.Word(
+                case="success",
+                startOffset=start_offset,
+                endOffset=end_offset,
+                word=display_word,
+                alignedWord=hyp_word,
+                phones=phones,
+                start=start,
+                end=end))
 
         elif op in ['insert', 'replace']:
-            out.append({
-                "case": "not-found-in-audio",
-                "startOffset": start_offset,
-                "endOffset": end_offset,
-                "word": display_word,
-            })
+            out.append(transcription.Word(
+                case="not-found-in-audio",
+                startOffset=start_offset,
+                endOffset=end_offset,
+                word=display_word))
     return out
 
 def word_diff(a, b):
diff --git a/gentle/forced_aligner.py b/gentle/forced_aligner.py
index b6c6658f..bff1c948 100644
--- a/gentle/forced_aligner.py
+++ b/gentle/forced_aligner.py
@@ -3,7 +3,8 @@
 from gentle import language_model
 from gentle import metasentence
 from gentle import multipass
-from gentle.transcription import MultiThreadedTranscriber, Transcription
+from gentle.transcriber import MultiThreadedTranscriber
+from gentle.transcription import Transcription
 
 class ForcedAligner():
 
@@ -31,7 +32,7 @@ def transcribe(self, wavfile, progress_cb=None, logging=None):
 
         # Perform a second-pass with unaligned words
         if logging is not None:
-            logging.info("%d unaligned words (of %d)" % (len([X for X in words if X.get("case") == "not-found-in-audio"]), len(words)))
+            logging.info("%d unaligned words (of %d)" % (len([X for X in words if X.case == "not-found-in-audio"]), len(words)))
 
         if progress_cb is not None:
             progress_cb({'status': 'ALIGNING'})
@@ -39,6 +40,6 @@ def transcribe(self, wavfile, progress_cb=None, logging=None):
         words = multipass.realign(wavfile, words, self.ms, resources=self.resources, nthreads=self.nthreads, progress_cb=progress_cb)
 
         if logging is not None:
-            logging.info("after 2nd pass: %d unaligned words (of %d)" % (len([X for X in words if X.get("case") == "not-found-in-audio"]), len(words)))
+            logging.info("after 2nd pass: %d unaligned words (of %d)" % (len([X for X in words if X.case == "not-found-in-audio"]), len(words)))
 
         return Transcription(words=words, transcript=self.transcript)
diff --git a/gentle/full_transcriber.py b/gentle/full_transcriber.py
index 025e7017..8d2bdfe8 100644
--- a/gentle/full_transcriber.py
+++ b/gentle/full_transcriber.py
@@ -1,7 +1,9 @@
 import os
 
 from gentle import kaldi_queue
-from gentle.transcription import MultiThreadedTranscriber, Transcription
+from gentle import transcription
+from gentle.transcriber import MultiThreadedTranscriber
+from gentle.transcription import Transcription
 
 class FullTranscriber():
 
@@ -24,17 +26,17 @@ def make_transcription_alignment(trans):
         transcript = ""
         words = []
         for t_wd in trans:
-            word = {
-                "case": "success",
-                "startOffset": len(transcript),
-                "endOffset": len(transcript) + len(t_wd["word"]),
-                "word": t_wd["word"],
-                "alignedWord": t_wd["word"],
-                "phones": t_wd["phones"],
-                "start": t_wd["start"],
-                "end": t_wd["start"] + t_wd["duration"]}
+            word = transcription.Word(
+                case="success",
+                startOffset=len(transcript),
+                endOffset=len(transcript) + len(t_wd.word),
+                word=t_wd.word,
+                alignedWord=t_wd.word,
+                phones=t_wd.phones,
+                start=t_wd.start,
+                end=t_wd.start + t_wd.duration)
             words.append(word)
 
-            transcript += word["word"] + " "
+            transcript += word.word + " "
 
         return Transcription(words=words, transcript=transcript)
diff --git a/gentle/multipass.py b/gentle/multipass.py
index 01f7d33f..052036eb 100644
--- a/gentle/multipass.py
+++ b/gentle/multipass.py
@@ -7,6 +7,7 @@
 from gentle import metasentence
 from gentle import language_model
 from gentle import diff_align
+from gentle import transcription
 
 def prepare_multipass(alignment):
     to_realign = []
@@ -14,9 +15,9 @@ def prepare_multipass(alignment):
     cur_unaligned_words = []
 
     for wd_idx,wd in enumerate(alignment):
-        if wd['case'] == 'not-found-in-audio':
+        if wd.case == 'not-found-in-audio':
             cur_unaligned_words.append(wd)
-        elif wd['case'] == 'success':
+        elif wd.case == 'success':
             if len(cur_unaligned_words) > 0:
                 to_realign.append({
                     "start": last_aligned_word,
@@ -41,12 +42,15 @@ def realign(wavfile, alignment, ms, resources, nthreads=4, progress_cb=None):
     def realign(chunk):
         wav_obj = wave.open(wavfile, 'r')
 
-        start_t = (chunk["start"] or {"end": 0})["end"]
-        end_t = chunk["end"]
-        if end_t is None:
+        if chunk["start"] is None:
+            start_t = 0
+        else:
+            start_t = chunk["start"].end
+
+        if chunk["end"] is None:
             end_t = wav_obj.getnframes() / float(wav_obj.getframerate())
         else:
-            end_t = end_t["start"]
+            end_t = chunk["end"].start
 
         duration = end_t - start_t
         if duration < 0.01 or duration > 60:
@@ -54,8 +58,8 @@ def realign(chunk):
             return
 
         # Create a language model
-        offset_offset = chunk['words'][0]['startOffset']
-        chunk_len = chunk['words'][-1]['endOffset'] - offset_offset
+        offset_offset = chunk['words'][0].startOffset
+        chunk_len = chunk['words'][-1].endOffset - offset_offset
         chunk_transcript = ms.raw_sentence[offset_offset:offset_offset+chunk_len].encode("utf-8")
         chunk_ms = metasentence.MetaSentence(chunk_transcript, resources.vocab)
         chunk_ks = chunk_ms.get_kaldi_sequence()
@@ -71,21 +75,21 @@ def realign(chunk):
         buf = wav_obj.readframes(int(duration * wav_obj.getframerate()))
 
         k.push_chunk(buf)
-        ret = k.get_final()
+        ret = [transcription.Word(**wd) for wd in k.get_final()]
         k.stop()
 
         word_alignment = diff_align.align(ret, chunk_ms)
 
         # Adjust startOffset, endOffset, and timing to match originals
         for wd in word_alignment:
-            if wd.get("end"):
+            if wd.end is not None:
                 # Apply timing offset
-                wd['start'] += start_t
-                wd['end'] += start_t
+                wd.start += start_t
+                wd.end += start_t
 
-            if wd.get("endOffset"):
-                wd['startOffset'] += offset_offset
-                wd['endOffset'] += offset_offset
+            if wd.endOffset is not None:
+                wd.startOffset += offset_offset
+                wd.endOffset += offset_offset
 
         # "chunk" should be replaced by "words"
         realignments.append({"chunk": chunk, "words": word_alignment})
diff --git a/gentle/standard_kaldi.py b/gentle/standard_kaldi.py
index 9ad91258..59bbd5cf 100644
--- a/gentle/standard_kaldi.py
+++ b/gentle/standard_kaldi.py
@@ -6,7 +6,6 @@
 import tempfile
 import wave
 
-from gentle import ffmpeg
 from util.paths import get_binary
 from gentle.rpc import RPCProtocol
 from gentle.resources import Resources
diff --git a/gentle/transcriber.py b/gentle/transcriber.py
new file mode 100644
index 00000000..a5fa52d7
--- /dev/null
+++ b/gentle/transcriber.py
@@ -0,0 +1,84 @@
+import math
+import logging
+import wave
+
+from gentle import transcription
+
+from multiprocessing.pool import ThreadPool as Pool
+
+class MultiThreadedTranscriber:
+    def __init__(self, kaldi_queue, chunk_len=20, overlap_t=2, nthreads=4):
+        self.chunk_len = chunk_len
+        self.overlap_t = overlap_t
+        self.nthreads = nthreads
+            
+        self.kaldi_queue = kaldi_queue
+
+    def transcribe(self, wavfile, progress_cb=None):
+        wav_obj = wave.open(wavfile, 'r')
+        duration = wav_obj.getnframes() / float(wav_obj.getframerate())
+        n_chunks = int(math.ceil(duration / float(self.chunk_len - self.overlap_t)))
+
+        chunks = []
+
+        def transcribe_chunk(idx):
+            wav_obj = wave.open(wavfile, 'r')
+            start_t = idx * (self.chunk_len - self.overlap_t)
+            # Seek
+            wav_obj.setpos(int(start_t * wav_obj.getframerate()))
+            # Read frames
+            buf = wav_obj.readframes(int(self.chunk_len * wav_obj.getframerate()))
+
+            k = self.kaldi_queue.get()
+            k.push_chunk(buf)
+            ret = k.get_final()
+            k.reset()
+            self.kaldi_queue.put(k)
+
+            chunks.append({"start": start_t, "words": ret})
+            logging.info('%d/%d' % (len(chunks), n_chunks))
+            if progress_cb is not None:
+                progress_cb({"message": ' '.join([X['word'] for X in ret]),
+                             "percent": len(chunks) / float(n_chunks)})
+
+
+        pool = Pool(min(n_chunks, self.nthreads))
+        pool.map(transcribe_chunk, range(n_chunks))
+        pool.close()
+        
+        chunks.sort(key=lambda x: x['start'])
+
+        # Combine chunks
+        # TODO: remove overlap? ...or just let the sequence aligner deal with it.
+        words = []
+        for c in chunks:
+            chunk_start = c['start']
+            for wd in c['words']:
+                wd['start'] += chunk_start
+                words.append(transcription.Word(**wd))
+
+        return words
+
+
+if __name__=='__main__':
+    # full transcription
+    from Queue import Queue
+    from util import ffmpeg
+    from gentle import standard_kaldi
+
+    import sys
+
+    import logging
+    logging.getLogger().setLevel('INFO')
+    
+    k_queue = Queue()
+    for i in range(3):
+        k_queue.put(standard_kaldi.Kaldi())
+
+    trans = MultiThreadedTranscriber(k_queue)
+
+    with gentle.resampled(sys.argv[1]) as filename:
+        out = trans.transcribe(filename)
+
+    open(sys.argv[2], 'w').write(out.to_json())
+
diff --git a/gentle/transcription.py b/gentle/transcription.py
index a130c9f1..ebf13648 100644
--- a/gentle/transcription.py
+++ b/gentle/transcription.py
@@ -1,64 +1,33 @@
 import csv
 import io
 import json
-import math
-import logging
-import wave
-
-from multiprocessing.pool import ThreadPool as Pool
-
-class MultiThreadedTranscriber:
-    def __init__(self, kaldi_queue, chunk_len=20, overlap_t=2, nthreads=4):
-        self.chunk_len = chunk_len
-        self.overlap_t = overlap_t
-        self.nthreads = nthreads
-            
-        self.kaldi_queue = kaldi_queue
-
-    def transcribe(self, wavfile, progress_cb=None):
-        wav_obj = wave.open(wavfile, 'r')
-        duration = wav_obj.getnframes() / float(wav_obj.getframerate())
-        n_chunks = int(math.ceil(duration / float(self.chunk_len - self.overlap_t)))
-
-        chunks = []
-
-        def transcribe_chunk(idx):
-            wav_obj = wave.open(wavfile, 'r')
-            start_t = idx * (self.chunk_len - self.overlap_t)
-            # Seek
-            wav_obj.setpos(int(start_t * wav_obj.getframerate()))
-            # Read frames
-            buf = wav_obj.readframes(int(self.chunk_len * wav_obj.getframerate()))
-
-            k = self.kaldi_queue.get()
-            k.push_chunk(buf)
-            ret = k.get_final()
-            k.reset()
-            self.kaldi_queue.put(k)
-
-            chunks.append({"start": start_t, "words": ret})
-            logging.info('%d/%d' % (len(chunks), n_chunks))
-            if progress_cb is not None:
-                progress_cb({"message": ' '.join([X['word'] for X in ret]),
-                             "percent": len(chunks) / float(n_chunks)})
-
-
-        pool = Pool(min(n_chunks, self.nthreads))
-        pool.map(transcribe_chunk, range(n_chunks))
-        pool.close()
-        
-        chunks.sort(key=lambda x: x['start'])
-
-        # Combine chunks
-        # TODO: remove overlap? ...or just let the sequence aligner deal with it.
-        words = []
-        for c in chunks:
-            chunk_start = c['start']
-            for wd in c['words']:
-                wd['start'] += chunk_start
-                words.append(wd)
-
-        return words
+
+from collections import defaultdict
+
+class Word:
+
+    def __init__(self, case=None, startOffset=None, endOffset=None, word=None, alignedWord=None, phones=None, start=None, end=None, duration=None):
+        self.case = case
+        self.startOffset = startOffset
+        self.endOffset = endOffset
+        self.word = word
+        self.alignedWord = alignedWord
+        self.phones = phones
+        self.start = start
+        self.end = end
+        self.duration = duration
+
+    def as_dict(self):
+        return { key:val for key, val in self.__dict__.iteritems() if val is not None }
+
+    def __eq__(self, other):
+        return self.__dict__ == other.__dict__
+
+    def __ne__(self, other):
+        return not self == other
+
+    def __repr__(self):
+        return "Word(" + " ".join(sorted([key + "=" + str(val) for key, val in self.as_dict().iteritems()])) + ")"
 
 class Transcription:
 
@@ -66,14 +35,37 @@ def __init__(self, transcript=None, words=None):
         self.transcript = transcript
         self.words = words
 
+    def __eq__(self, other):
+        return self.transcript == other.transcript and self.words == other.words
+
     def to_json(self, **kwargs):
         '''Return a JSON representation of the aligned transcript'''
+        options = {
+                'sort_keys':    True,
+                'indent':       4,
+                'separators':   (',', ': '),
+                }
+        options.update(kwargs)
+
         container = {}
         if self.transcript:
             container['transcript'] = self.transcript
         if self.words: 
-            container['words'] = self.words
-        return json.dumps(container, **kwargs)
+            container['words'] = [word.as_dict() for word in self.words]
+        return json.dumps(container, **options)
+
+    @classmethod
+    def from_json(cls, json_str):
+        return cls._from_jsondata(json.loads(json_str))
+
+    @classmethod
+    def from_jsonfile(cls, filename):
+        with open(filename) as fh:
+            return cls._from_jsondata(json.load(fh))
+
+    @classmethod
+    def _from_jsondata(cls, data):
+        return cls(transcript = data['transcript'], words = [Word(**wd) for wd in data['words']])
 
     def to_csv(self):
         '''Return a CSV representation of the aligned transcript. Format:
@@ -84,34 +76,24 @@ def to_csv(self):
         buf = io.BytesIO()
         w = csv.writer(buf)
         for X in self.words:
-            if X.get("case") not in ("success", "not-found-in-audio"):
+            if X.case not in ("success", "not-found-in-audio"):
                 continue
-            row = [X["word"],
-                X.get("alignedWord"),
-                X.get("start"),
-                X.get("end")
+            row = [X.word,
+                X.alignedWord,
+                X.start,
+                X.end
             ]
             w.writerow(row)
         return buf.getvalue()
 
-if __name__=='__main__':
-    # full transcription
-    from Queue import Queue
-    from util import ffmpeg
-    from gentle import standard_kaldi
-
-    import sys
-
-    import logging
-    logging.getLogger().setLevel('INFO')
-    
-    k_queue = Queue()
-    for i in range(3):
-        k_queue.put(standard_kaldi.Kaldi())
-
-    trans = MultiThreadedTranscriber(k_queue)
-
-    with gentle.resampled(sys.argv[1]) as filename:
-        out = trans.transcribe(filename)
-
-    open(sys.argv[2], 'w').write(out.to_json())
+    def stats(self):
+        counts = defaultdict(int)
+        for word in self.words:
+            counts[word.case] += 1
+        stats = {}
+        stats['total'] = len(self.words)
+        for key, val in counts.iteritems():
+            stats[key] = val
+        return stats
+
+Transcription.Word = Word
diff --git a/pytest.ini b/pytest.ini
new file mode 100644
index 00000000..21ffa1d4
--- /dev/null
+++ b/pytest.ini
@@ -0,0 +1,2 @@
+[pytest]
+addopts = --disable-pytest-warnings
diff --git a/tests/.gitignore b/tests/.gitignore
new file mode 100644
index 00000000..b5f1fe66
--- /dev/null
+++ b/tests/.gitignore
@@ -0,0 +1,2 @@
+!data/
+data/tmp
diff --git a/tests/__init__.py b/tests/__init__.py
new file mode 100644
index 00000000..56764ff1
--- /dev/null
+++ b/tests/__init__.py
@@ -0,0 +1 @@
+# nothing here right now
diff --git a/tests/data/expected/forced-harvard-sentences-list30.json b/tests/data/expected/forced-harvard-sentences-list30.json
new file mode 100644
index 00000000..eb87a521
--- /dev/null
+++ b/tests/data/expected/forced-harvard-sentences-list30.json
@@ -0,0 +1,2015 @@
+{
+  "transcript": "The mute muffled the high tones of the horn.\nThe gold ring fits only a pierced ear.\nThe old pan was covered with hard fudge.\nWatch the log float in the wide river.\nThe node on the stalk of wheat grew daily.\nThe heap of fallen leaves was set on fire.\nWrite fast, if you want to finish early.\nHis shirt was clean but one button was gone.\nThe barrel of beer was a brew of malt and hops.\nTin cans are absent from store shelves.\n",
+  "words": [
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 1.29,
+      "endOffset": 3,
+      "phones": [
+        {
+          "duration": 0.18,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.08,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 1.03,
+      "startOffset": 0,
+      "word": "The"
+    },
+    {
+      "alignedWord": "mute",
+      "case": "success",
+      "end": 1.6400000000000001,
+      "endOffset": 8,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "m_B"
+        },
+        {
+          "duration": 0.12,
+          "phone": "y_I"
+        },
+        {
+          "duration": 0.11,
+          "phone": "uw_I"
+        },
+        {
+          "duration": 0.04,
+          "phone": "t_E"
+        }
+      ],
+      "start": 1.29,
+      "startOffset": 4,
+      "word": "mute"
+    },
+    {
+      "alignedWord": "muffled",
+      "case": "success",
+      "end": 2.42,
+      "endOffset": 16,
+      "phones": [
+        {
+          "duration": 0.23,
+          "phone": "m_B"
+        },
+        {
+          "duration": 0.1,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.14,
+          "phone": "f_I"
+        },
+        {
+          "duration": 0.04,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.13,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.07,
+          "phone": "d_E"
+        }
+      ],
+      "start": 1.71,
+      "startOffset": 9,
+      "word": "muffled"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 2.6700000000000004,
+      "endOffset": 20,
+      "phones": [
+        {
+          "duration": 0.13,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.09,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 2.45,
+      "startOffset": 17,
+      "word": "the"
+    },
+    {
+      "alignedWord": "high",
+      "case": "success",
+      "end": 2.83,
+      "endOffset": 25,
+      "phones": [
+        {
+          "duration": 0.11,
+          "phone": "hh_B"
+        },
+        {
+          "duration": 0.05,
+          "phone": "ay_E"
+        }
+      ],
+      "start": 2.67,
+      "startOffset": 21,
+      "word": "high"
+    },
+    {
+      "alignedWord": "tones",
+      "case": "success",
+      "end": 3.55,
+      "endOffset": 31,
+      "phones": [
+        {
+          "duration": 0.14,
+          "phone": "t_B"
+        },
+        {
+          "duration": 0.25,
+          "phone": "ow_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "n_I"
+        },
+        {
+          "duration": 0.07,
+          "phone": "z_E"
+        }
+      ],
+      "start": 3.01,
+      "startOffset": 26,
+      "word": "tones"
+    },
+    {
+      "alignedWord": "of",
+      "case": "success",
+      "end": 3.74,
+      "endOffset": 34,
+      "phones": [
+        {
+          "duration": 0.09,
+          "phone": "ah_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "v_E"
+        }
+      ],
+      "start": 3.58,
+      "startOffset": 32,
+      "word": "of"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 3.8400000000000003,
+      "endOffset": 38,
+      "phones": [
+        {
+          "duration": 0.03,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 3.74,
+      "startOffset": 35,
+      "word": "the"
+    },
+    {
+      "alignedWord": "horn",
+      "case": "success",
+      "end": 4.34,
+      "endOffset": 43,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "hh_B"
+        },
+        {
+          "duration": 0.18,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.12,
+          "phone": "r_I"
+        },
+        {
+          "duration": 0.12,
+          "phone": "n_E"
+        }
+      ],
+      "start": 3.84,
+      "startOffset": 39,
+      "word": "horn"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 6.36,
+      "endOffset": 48,
+      "phones": [
+        {
+          "duration": 0.11,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.08,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 6.17,
+      "startOffset": 45,
+      "word": "The"
+    },
+    {
+      "alignedWord": "gold",
+      "case": "success",
+      "end": 6.8500000000000005,
+      "endOffset": 53,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "g_B"
+        },
+        {
+          "duration": 0.11,
+          "phone": "ow_I"
+        },
+        {
+          "duration": 0.18,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.12,
+          "phone": "d_E"
+        }
+      ],
+      "start": 6.36,
+      "startOffset": 49,
+      "word": "gold"
+    },
+    {
+      "alignedWord": "ring",
+      "case": "success",
+      "end": 7.27,
+      "endOffset": 58,
+      "phones": [
+        {
+          "duration": 0.11,
+          "phone": "r_B"
+        },
+        {
+          "duration": 0.09,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.22,
+          "phone": "ng_E"
+        }
+      ],
+      "start": 6.85,
+      "startOffset": 54,
+      "word": "ring"
+    },
+    {
+      "alignedWord": "fits",
+      "case": "success",
+      "end": 7.92,
+      "endOffset": 63,
+      "phones": [
+        {
+          "duration": 0.22,
+          "phone": "f_B"
+        },
+        {
+          "duration": 0.12,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.1,
+          "phone": "t_I"
+        },
+        {
+          "duration": 0.18,
+          "phone": "s_E"
+        }
+      ],
+      "start": 7.3,
+      "startOffset": 59,
+      "word": "fits"
+    },
+    {
+      "alignedWord": "only",
+      "case": "success",
+      "end": 8.63,
+      "endOffset": 68,
+      "phones": [
+        {
+          "duration": 0.11,
+          "phone": "ow_B"
+        },
+        {
+          "duration": 0.25,
+          "phone": "n_I"
+        },
+        {
+          "duration": 0.09,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.12,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 8.06,
+      "startOffset": 64,
+      "word": "only"
+    },
+    {
+      "alignedWord": "a",
+      "case": "success",
+      "end": 8.73,
+      "endOffset": 70,
+      "phones": [
+        {
+          "duration": 0.1,
+          "phone": "ah_S"
+        }
+      ],
+      "start": 8.63,
+      "startOffset": 69,
+      "word": "a"
+    },
+    {
+      "alignedWord": "pierced",
+      "case": "success",
+      "end": 9.27,
+      "endOffset": 78,
+      "phones": [
+        {
+          "duration": 0.09,
+          "phone": "p_B"
+        },
+        {
+          "duration": 0.12,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.1,
+          "phone": "r_I"
+        },
+        {
+          "duration": 0.04,
+          "phone": "s_I"
+        },
+        {
+          "duration": 0.14,
+          "phone": "t_E"
+        }
+      ],
+      "start": 8.78,
+      "startOffset": 71,
+      "word": "pierced"
+    },
+    {
+      "alignedWord": "ear",
+      "case": "success",
+      "end": 9.74,
+      "endOffset": 82,
+      "phones": [
+        {
+          "duration": 0.23,
+          "phone": "ih_B"
+        },
+        {
+          "duration": 0.21,
+          "phone": "r_E"
+        }
+      ],
+      "start": 9.3,
+      "startOffset": 79,
+      "word": "ear"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 12.11,
+      "endOffset": 87,
+      "phones": [
+        {
+          "duration": 0.16,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.11,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 11.84,
+      "startOffset": 84,
+      "word": "The"
+    },
+    {
+      "alignedWord": "old",
+      "case": "success",
+      "end": 12.450000000000001,
+      "endOffset": 91,
+      "phones": [
+        {
+          "duration": 0.19,
+          "phone": "ow_B"
+        },
+        {
+          "duration": 0.06,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.06,
+          "phone": "d_E"
+        }
+      ],
+      "start": 12.14,
+      "startOffset": 88,
+      "word": "old"
+    },
+    {
+      "alignedWord": "pan",
+      "case": "success",
+      "end": 13.11,
+      "endOffset": 95,
+      "phones": [
+        {
+          "duration": 0.17,
+          "phone": "p_B"
+        },
+        {
+          "duration": 0.26,
+          "phone": "ae_I"
+        },
+        {
+          "duration": 0.17,
+          "phone": "n_E"
+        }
+      ],
+      "start": 12.51,
+      "startOffset": 92,
+      "word": "pan"
+    },
+    {
+      "alignedWord": "was",
+      "case": "success",
+      "end": 13.299999999999999,
+      "endOffset": 99,
+      "phones": [
+        {
+          "duration": 0.06,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.05,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "z_E"
+        }
+      ],
+      "start": 13.11,
+      "startOffset": 96,
+      "word": "was"
+    },
+    {
+      "alignedWord": "covered",
+      "case": "success",
+      "end": 13.82,
+      "endOffset": 107,
+      "phones": [
+        {
+          "duration": 0.17,
+          "phone": "k_B"
+        },
+        {
+          "duration": 0.09,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "v_I"
+        },
+        {
+          "duration": 0.09,
+          "phone": "er_I"
+        },
+        {
+          "duration": 0.09,
+          "phone": "d_E"
+        }
+      ],
+      "start": 13.3,
+      "startOffset": 100,
+      "word": "covered"
+    },
+    {
+      "alignedWord": "with",
+      "case": "success",
+      "end": 14.030000000000001,
+      "endOffset": 112,
+      "phones": [
+        {
+          "duration": 0.09,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.06,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.06,
+          "phone": "th_E"
+        }
+      ],
+      "start": 13.82,
+      "startOffset": 108,
+      "word": "with"
+    },
+    {
+      "alignedWord": "hard",
+      "case": "success",
+      "end": 14.58,
+      "endOffset": 117,
+      "phones": [
+        {
+          "duration": 0.22,
+          "phone": "hh_B"
+        },
+        {
+          "duration": 0.13,
+          "phone": "aa_I"
+        },
+        {
+          "duration": 0.09,
+          "phone": "r_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "d_E"
+        }
+      ],
+      "start": 14.06,
+      "startOffset": 113,
+      "word": "hard"
+    },
+    {
+      "alignedWord": "fudge",
+      "case": "success",
+      "end": 15.42,
+      "endOffset": 123,
+      "phones": [
+        {
+          "duration": 0.03,
+          "phone": "f_B"
+        },
+        {
+          "duration": 0.21,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.38,
+          "phone": "jh_E"
+        }
+      ],
+      "start": 14.8,
+      "startOffset": 118,
+      "word": "fudge"
+    },
+    {
+      "alignedWord": "watch",
+      "case": "success",
+      "end": 17.98,
+      "endOffset": 130,
+      "phones": [
+        {
+          "duration": 0.18,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.19,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.17,
+          "phone": "ch_E"
+        }
+      ],
+      "start": 17.44,
+      "startOffset": 125,
+      "word": "Watch"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 18.19,
+      "endOffset": 134,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.11,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 18,
+      "startOffset": 131,
+      "word": "the"
+    },
+    {
+      "alignedWord": "log",
+      "case": "success",
+      "end": 18.64,
+      "endOffset": 138,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "l_B"
+        },
+        {
+          "duration": 0.24,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.13,
+          "phone": "g_E"
+        }
+      ],
+      "start": 18.19,
+      "startOffset": 135,
+      "word": "log"
+    },
+    {
+      "alignedWord": "float",
+      "case": "success",
+      "end": 19.21,
+      "endOffset": 144,
+      "phones": [
+        {
+          "duration": 0.09,
+          "phone": "f_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.17,
+          "phone": "ow_I"
+        },
+        {
+          "duration": 0.11,
+          "phone": "t_E"
+        }
+      ],
+      "start": 18.77,
+      "startOffset": 139,
+      "word": "float"
+    },
+    {
+      "alignedWord": "in",
+      "case": "success",
+      "end": 19.48,
+      "endOffset": 147,
+      "phones": [
+        {
+          "duration": 0.1,
+          "phone": "ih_B"
+        },
+        {
+          "duration": 0.1,
+          "phone": "n_E"
+        }
+      ],
+      "start": 19.28,
+      "startOffset": 145,
+      "word": "in"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 19.6,
+      "endOffset": 151,
+      "phones": [
+        {
+          "duration": 0.04,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.08,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 19.48,
+      "startOffset": 148,
+      "word": "the"
+    },
+    {
+      "alignedWord": "wide",
+      "case": "success",
+      "end": 20.1,
+      "endOffset": 156,
+      "phones": [
+        {
+          "duration": 0.15,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.29,
+          "phone": "ay_I"
+        },
+        {
+          "duration": 0.06,
+          "phone": "d_E"
+        }
+      ],
+      "start": 19.6,
+      "startOffset": 152,
+      "word": "wide"
+    },
+    {
+      "alignedWord": "river",
+      "case": "success",
+      "end": 20.779999999999998,
+      "endOffset": 162,
+      "phones": [
+        {
+          "duration": 0.23,
+          "phone": "r_B"
+        },
+        {
+          "duration": 0.05,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.1,
+          "phone": "v_I"
+        },
+        {
+          "duration": 0.27,
+          "phone": "er_E"
+        }
+      ],
+      "start": 20.13,
+      "startOffset": 157,
+      "word": "river"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 22.9,
+      "endOffset": 167,
+      "phones": [
+        {
+          "duration": 0.18,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.09,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 22.63,
+      "startOffset": 164,
+      "word": "The"
+    },
+    {
+      "alignedWord": "node",
+      "case": "success",
+      "end": 23.389999999999997,
+      "endOffset": 172,
+      "phones": [
+        {
+          "duration": 0.11,
+          "phone": "n_B"
+        },
+        {
+          "duration": 0.3,
+          "phone": "ow_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "d_E"
+        }
+      ],
+      "start": 22.9,
+      "startOffset": 168,
+      "word": "node"
+    },
+    {
+      "alignedWord": "on",
+      "case": "success",
+      "end": 23.61,
+      "endOffset": 175,
+      "phones": [
+        {
+          "duration": 0.13,
+          "phone": "ao_B"
+        },
+        {
+          "duration": 0.09,
+          "phone": "n_E"
+        }
+      ],
+      "start": 23.39,
+      "startOffset": 173,
+      "word": "on"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 23.72,
+      "endOffset": 179,
+      "phones": [
+        {
+          "duration": 0.04,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 23.61,
+      "startOffset": 176,
+      "word": "the"
+    },
+    {
+      "alignedWord": "stalk",
+      "case": "success",
+      "end": 24.369999999999997,
+      "endOffset": 185,
+      "phones": [
+        {
+          "duration": 0.17,
+          "phone": "s_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "t_I"
+        },
+        {
+          "duration": 0.27,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.14,
+          "phone": "k_E"
+        }
+      ],
+      "start": 23.72,
+      "startOffset": 180,
+      "word": "stalk"
+    },
+    {
+      "alignedWord": "of",
+      "case": "success",
+      "end": 24.549999999999997,
+      "endOffset": 188,
+      "phones": [
+        {
+          "duration": 0.09,
+          "phone": "ah_B"
+        },
+        {
+          "duration": 0.06,
+          "phone": "v_E"
+        }
+      ],
+      "start": 24.4,
+      "startOffset": 186,
+      "word": "of"
+    },
+    {
+      "alignedWord": "wheat",
+      "case": "success",
+      "end": 24.990000000000002,
+      "endOffset": 194,
+      "phones": [
+        {
+          "duration": 0.03,
+          "phone": "hh_B"
+        },
+        {
+          "duration": 0.15,
+          "phone": "w_I"
+        },
+        {
+          "duration": 0.15,
+          "phone": "iy_I"
+        },
+        {
+          "duration": 0.11,
+          "phone": "t_E"
+        }
+      ],
+      "start": 24.55,
+      "startOffset": 189,
+      "word": "wheat"
+    },
+    {
+      "alignedWord": "grew",
+      "case": "success",
+      "end": 25.479999999999997,
+      "endOffset": 199,
+      "phones": [
+        {
+          "duration": 0.16,
+          "phone": "g_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "r_I"
+        },
+        {
+          "duration": 0.17,
+          "phone": "uw_E"
+        }
+      ],
+      "start": 25.08,
+      "startOffset": 195,
+      "word": "grew"
+    },
+    {
+      "alignedWord": "daily",
+      "case": "success",
+      "end": 26.14,
+      "endOffset": 205,
+      "phones": [
+        {
+          "duration": 0.11,
+          "phone": "d_B"
+        },
+        {
+          "duration": 0.2,
+          "phone": "ey_I"
+        },
+        {
+          "duration": 0.07,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.28,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 25.48,
+      "startOffset": 200,
+      "word": "daily"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 27.81,
+      "endOffset": 210,
+      "phones": [
+        {
+          "duration": 0.19,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.03,
+          "phone": "ah_E"
+        }
+      ],
+      "start": 27.59,
+      "startOffset": 207,
+      "word": "The"
+    },
+    {
+      "alignedWord": "heap",
+      "case": "success",
+      "end": 28.240000000000002,
+      "endOffset": 215,
+      "phones": [
+        {
+          "duration": 0.22,
+          "phone": "hh_B"
+        },
+        {
+          "duration": 0.16,
+          "phone": "iy_I"
+        },
+        {
+          "duration": 0.05,
+          "phone": "p_E"
+        }
+      ],
+      "start": 27.810000000000002,
+      "startOffset": 211,
+      "word": "heap"
+    },
+    {
+      "alignedWord": "of",
+      "case": "success",
+      "end": 28.5,
+      "endOffset": 218,
+      "phones": [
+        {
+          "duration": 0.13,
+          "phone": "ah_B"
+        },
+        {
+          "duration": 0.08,
+          "phone": "v_E"
+        }
+      ],
+      "start": 28.29,
+      "startOffset": 216,
+      "word": "of"
+    },
+    {
+      "alignedWord": "fallen",
+      "case": "success",
+      "end": 29.05,
+      "endOffset": 225,
+      "phones": [
+        {
+          "duration": 0.09,
+          "phone": "f_B"
+        },
+        {
+          "duration": 0.13,
+          "phone": "aa_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.04,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.15,
+          "phone": "n_E"
+        }
+      ],
+      "start": 28.560000000000002,
+      "startOffset": 219,
+      "word": "fallen"
+    },
+    {
+      "alignedWord": "leaves",
+      "case": "success",
+      "end": 29.580000000000002,
+      "endOffset": 232,
+      "phones": [
+        {
+          "duration": 0.07,
+          "phone": "l_B"
+        },
+        {
+          "duration": 0.25,
+          "phone": "iy_I"
+        },
+        {
+          "duration": 0.04,
+          "phone": "v_I"
+        },
+        {
+          "duration": 0.17,
+          "phone": "z_E"
+        }
+      ],
+      "start": 29.05,
+      "startOffset": 226,
+      "word": "leaves"
+    },
+    {
+      "alignedWord": "was",
+      "case": "success",
+      "end": 29.89,
+      "endOffset": 236,
+      "phones": [
+        {
+          "duration": 0.13,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.06,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.06,
+          "phone": "z_E"
+        }
+      ],
+      "start": 29.64,
+      "startOffset": 233,
+      "word": "was"
+    },
+    {
+      "alignedWord": "set",
+      "case": "success",
+      "end": 30.26,
+      "endOffset": 240,
+      "phones": [
+        {
+          "duration": 0.14,
+          "phone": "s_B"
+        },
+        {
+          "duration": 0.12,
+          "phone": "eh_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "t_E"
+        }
+      ],
+      "start": 29.92,
+      "startOffset": 237,
+      "word": "set"
+    },
+    {
+      "alignedWord": "on",
+      "case": "success",
+      "end": 30.52,
+      "endOffset": 243,
+      "phones": [
+        {
+          "duration": 0.12,
+          "phone": "ao_B"
+        },
+        {
+          "duration": 0.1,
+          "phone": "n_E"
+        }
+      ],
+      "start": 30.3,
+      "startOffset": 241,
+      "word": "on"
+    },
+    {
+      "alignedWord": "fire",
+      "case": "success",
+      "end": 31.21,
+      "endOffset": 248,
+      "phones": [
+        {
+          "duration": 0.18,
+          "phone": "f_B"
+        },
+        {
+          "duration": 0.25,
+          "phone": "ay_I"
+        },
+        {
+          "duration": 0.26,
+          "phone": "r_E"
+        }
+      ],
+      "start": 30.52,
+      "startOffset": 244,
+      "word": "fire"
+    },
+    {
+      "alignedWord": "write",
+      "case": "success",
+      "end": 33.22,
+      "endOffset": 255,
+      "phones": [
+        {
+          "duration": 0.19,
+          "phone": "r_B"
+        },
+        {
+          "duration": 0.08,
+          "phone": "ay_I"
+        },
+        {
+          "duration": 0.1,
+          "phone": "t_E"
+        }
+      ],
+      "start": 32.85,
+      "startOffset": 250,
+      "word": "Write"
+    },
+    {
+      "alignedWord": "fast",
+      "case": "success",
+      "end": 33.79,
+      "endOffset": 260,
+      "phones": [
+        {
+          "duration": 0.14,
+          "phone": "f_B"
+        },
+        {
+          "duration": 0.24,
+          "phone": "ae_I"
+        },
+        {
+          "duration": 0.06,
+          "phone": "s_I"
+        },
+        {
+          "duration": 0.09,
+          "phone": "t_E"
+        }
+      ],
+      "start": 33.26,
+      "startOffset": 256,
+      "word": "fast"
+    },
+    {
+      "alignedWord": "if",
+      "case": "success",
+      "end": 33.96,
+      "endOffset": 264,
+      "phones": [
+        {
+          "duration": 0.09,
+          "phone": "ih_B"
+        },
+        {
+          "duration": 0.08,
+          "phone": "f_E"
+        }
+      ],
+      "start": 33.79,
+      "startOffset": 262,
+      "word": "if"
+    },
+    {
+      "alignedWord": "you",
+      "case": "success",
+      "end": 34.08,
+      "endOffset": 268,
+      "phones": [
+        {
+          "duration": 0.06,
+          "phone": "y_B"
+        },
+        {
+          "duration": 0.06,
+          "phone": "uw_E"
+        }
+      ],
+      "start": 33.96,
+      "startOffset": 265,
+      "word": "you"
+    },
+    {
+      "alignedWord": "want",
+      "case": "success",
+      "end": 34.379999999999995,
+      "endOffset": 273,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.1,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.05,
+          "phone": "n_I"
+        },
+        {
+          "duration": 0.07,
+          "phone": "t_E"
+        }
+      ],
+      "start": 34.08,
+      "startOffset": 269,
+      "word": "want"
+    },
+    {
+      "alignedWord": "to",
+      "case": "success",
+      "end": 34.489999999999995,
+      "endOffset": 276,
+      "phones": [
+        {
+          "duration": 0.04,
+          "phone": "t_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "uw_E"
+        }
+      ],
+      "start": 34.379999999999995,
+      "startOffset": 274,
+      "word": "to"
+    },
+    {
+      "alignedWord": "finish",
+      "case": "success",
+      "end": 35.019999999999996,
+      "endOffset": 283,
+      "phones": [
+        {
+          "duration": 0.11,
+          "phone": "f_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.06,
+          "phone": "n_I"
+        },
+        {
+          "duration": 0.07,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.22,
+          "phone": "sh_E"
+        }
+      ],
+      "start": 34.489999999999995,
+      "startOffset": 277,
+      "word": "finish"
+    },
+    {
+      "alignedWord": "early",
+      "case": "success",
+      "end": 35.730000000000004,
+      "endOffset": 289,
+      "phones": [
+        {
+          "duration": 0.28,
+          "phone": "er_B"
+        },
+        {
+          "duration": 0.13,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.26,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 35.06,
+      "startOffset": 284,
+      "word": "early"
+    },
+    {
+      "alignedWord": "his",
+      "case": "success",
+      "end": 37.8,
+      "endOffset": 294,
+      "phones": [
+        {
+          "duration": 0.14,
+          "phone": "hh_B"
+        },
+        {
+          "duration": 0.06,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "z_E"
+        }
+      ],
+      "start": 37.519999999999996,
+      "startOffset": 291,
+      "word": "His"
+    },
+    {
+      "alignedWord": "shirt",
+      "case": "success",
+      "end": 37.879999999999995,
+      "endOffset": 300,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "sh_B"
+        }
+      ],
+      "start": 37.8,
+      "startOffset": 295,
+      "word": "shirt"
+    },
+    {
+      "alignedWord": "was",
+      "case": "success",
+      "end": 38.459999999999994,
+      "endOffset": 304,
+      "phones": [
+        {
+          "duration": 0.12,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.05,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.07,
+          "phone": "z_E"
+        }
+      ],
+      "start": 38.22,
+      "startOffset": 301,
+      "word": "was"
+    },
+    {
+      "alignedWord": "clean",
+      "case": "success",
+      "end": 39.05,
+      "endOffset": 310,
+      "phones": [
+        {
+          "duration": 0.17,
+          "phone": "k_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.24,
+          "phone": "iy_I"
+        },
+        {
+          "duration": 0.11,
+          "phone": "n_E"
+        }
+      ],
+      "start": 38.459999999999994,
+      "startOffset": 305,
+      "word": "clean"
+    },
+    {
+      "alignedWord": "but",
+      "case": "success",
+      "end": 39.309999999999995,
+      "endOffset": 314,
+      "phones": [
+        {
+          "duration": 0.12,
+          "phone": "b_B"
+        },
+        {
+          "duration": 0.03,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "t_E"
+        }
+      ],
+      "start": 39.08,
+      "startOffset": 311,
+      "word": "but"
+    },
+    {
+      "alignedWord": "one",
+      "case": "success",
+      "end": 39.629999999999995,
+      "endOffset": 318,
+      "phones": [
+        {
+          "duration": 0.14,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.07,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.11,
+          "phone": "n_E"
+        }
+      ],
+      "start": 39.309999999999995,
+      "startOffset": 315,
+      "word": "one"
+    },
+    {
+      "alignedWord": "button",
+      "case": "success",
+      "end": 39.879999999999995,
+      "endOffset": 325,
+      "phones": [
+        {
+          "duration": 0.05,
+          "phone": "b_B"
+        },
+        {
+          "duration": 0.09,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.06,
+          "phone": "t_I"
+        },
+        {
+          "duration": 0.05,
+          "phone": "ah_I"
+        }
+      ],
+      "start": 39.629999999999995,
+      "startOffset": 319,
+      "word": "button"
+    },
+    {
+      "alignedWord": "was",
+      "case": "success",
+      "end": 40.18,
+      "endOffset": 329,
+      "phones": [
+        {
+          "duration": 0.06,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.04,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "z_E"
+        }
+      ],
+      "start": 40,
+      "startOffset": 326,
+      "word": "was"
+    },
+    {
+      "alignedWord": "gone",
+      "case": "success",
+      "end": 40.76,
+      "endOffset": 334,
+      "phones": [
+        {
+          "duration": 0.1,
+          "phone": "g_B"
+        },
+        {
+          "duration": 0.3,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.18,
+          "phone": "n_E"
+        }
+      ],
+      "start": 40.18,
+      "startOffset": 330,
+      "word": "gone"
+    },
+    {
+      "alignedWord": "the",
+      "case": "success",
+      "end": 42.82,
+      "endOffset": 339,
+      "phones": [
+        {
+          "duration": 0.17,
+          "phone": "dh_B"
+        },
+        {
+          "duration": 0.08,
+          "phone": "iy_E"
+        }
+      ],
+      "start": 42.57,
+      "startOffset": 336,
+      "word": "The"
+    },
+    {
+      "alignedWord": "barrel",
+      "case": "success",
+      "end": 43.34,
+      "endOffset": 346,
+      "phones": [
+        {
+          "duration": 0.1,
+          "phone": "b_B"
+        },
+        {
+          "duration": 0.17,
+          "phone": "eh_I"
+        },
+        {
+          "duration": 0.07,
+          "phone": "r_I"
+        },
+        {
+          "duration": 0.1,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "l_E"
+        }
+      ],
+      "start": 42.82,
+      "startOffset": 340,
+      "word": "barrel"
+    },
+    {
+      "alignedWord": "of",
+      "case": "success",
+      "end": 43.480000000000004,
+      "endOffset": 349,
+      "phones": [
+        {
+          "duration": 0.03,
+          "phone": "ah_B"
+        },
+        {
+          "duration": 0.11,
+          "phone": "v_E"
+        }
+      ],
+      "start": 43.34,
+      "startOffset": 347,
+      "word": "of"
+    },
+    {
+      "alignedWord": "beer",
+      "case": "success",
+      "end": 44.04,
+      "endOffset": 354,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "b_B"
+        },
+        {
+          "duration": 0.25,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.2,
+          "phone": "r_E"
+        }
+      ],
+      "start": 43.51,
+      "startOffset": 350,
+      "word": "beer"
+    },
+    {
+      "alignedWord": "was",
+      "case": "success",
+      "end": 44.27,
+      "endOffset": 358,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "w_B"
+        },
+        {
+          "duration": 0.04,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "z_E"
+        }
+      ],
+      "start": 44.07,
+      "startOffset": 355,
+      "word": "was"
+    },
+    {
+      "alignedWord": "a",
+      "case": "success",
+      "end": 44.339999999999996,
+      "endOffset": 360,
+      "phones": [
+        {
+          "duration": 0.07,
+          "phone": "ah_S"
+        }
+      ],
+      "start": 44.269999999999996,
+      "startOffset": 359,
+      "word": "a"
+    },
+    {
+      "alignedWord": "brew",
+      "case": "success",
+      "end": 44.78,
+      "endOffset": 365,
+      "phones": [
+        {
+          "duration": 0.13,
+          "phone": "b_B"
+        },
+        {
+          "duration": 0.03,
+          "phone": "r_I"
+        },
+        {
+          "duration": 0.28,
+          "phone": "uw_E"
+        }
+      ],
+      "start": 44.34,
+      "startOffset": 361,
+      "word": "brew"
+    },
+    {
+      "alignedWord": "of",
+      "case": "success",
+      "end": 44.95,
+      "endOffset": 368,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "ah_B"
+        },
+        {
+          "duration": 0.09,
+          "phone": "v_E"
+        }
+      ],
+      "start": 44.78,
+      "startOffset": 366,
+      "word": "of"
+    },
+    {
+      "alignedWord": "malt",
+      "case": "success",
+      "end": 45.45,
+      "endOffset": 373,
+      "phones": [
+        {
+          "duration": 0.14,
+          "phone": "m_B"
+        },
+        {
+          "duration": 0.12,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.18,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.03,
+          "phone": "t_E"
+        }
+      ],
+      "start": 44.980000000000004,
+      "startOffset": 369,
+      "word": "malt"
+    },
+    {
+      "alignedWord": "and",
+      "case": "success",
+      "end": 45.78,
+      "endOffset": 377,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "ah_B"
+        },
+        {
+          "duration": 0.05,
+          "phone": "n_I"
+        },
+        {
+          "duration": 0.05,
+          "phone": "d_E"
+        }
+      ],
+      "start": 45.6,
+      "startOffset": 374,
+      "word": "and"
+    },
+    {
+      "alignedWord": "hops",
+      "case": "success",
+      "end": 46.32,
+      "endOffset": 382,
+      "phones": [
+        {
+          "duration": 0.12,
+          "phone": "hh_B"
+        },
+        {
+          "duration": 0.2,
+          "phone": "aa_I"
+        },
+        {
+          "duration": 0.14,
+          "phone": "p_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "s_E"
+        }
+      ],
+      "start": 45.78,
+      "startOffset": 378,
+      "word": "hops"
+    },
+    {
+      "alignedWord": "tin",
+      "case": "success",
+      "end": 48.93,
+      "endOffset": 387,
+      "phones": [
+        {
+          "duration": 0.11,
+          "phone": "t_B"
+        },
+        {
+          "duration": 0.12,
+          "phone": "ih_I"
+        },
+        {
+          "duration": 0.12,
+          "phone": "n_E"
+        }
+      ],
+      "start": 48.58,
+      "startOffset": 384,
+      "word": "Tin"
+    },
+    {
+      "alignedWord": "cans",
+      "case": "success",
+      "end": 49.49,
+      "endOffset": 392,
+      "phones": [
+        {
+          "duration": 0.1,
+          "phone": "k_B"
+        },
+        {
+          "duration": 0.27,
+          "phone": "ae_I"
+        },
+        {
+          "duration": 0.06,
+          "phone": "n_I"
+        },
+        {
+          "duration": 0.13,
+          "phone": "z_E"
+        }
+      ],
+      "start": 48.93,
+      "startOffset": 388,
+      "word": "cans"
+    },
+    {
+      "alignedWord": "are",
+      "case": "success",
+      "end": 49.63,
+      "endOffset": 396,
+      "phones": [
+        {
+          "duration": 0.14,
+          "phone": "er_S"
+        }
+      ],
+      "start": 49.49,
+      "startOffset": 393,
+      "word": "are"
+    },
+    {
+      "alignedWord": "absent",
+      "case": "success",
+      "end": 50.220000000000006,
+      "endOffset": 403,
+      "phones": [
+        {
+          "duration": 0.24,
+          "phone": "ae_B"
+        },
+        {
+          "duration": 0.1,
+          "phone": "b_I"
+        },
+        {
+          "duration": 0.07,
+          "phone": "s_I"
+        },
+        {
+          "duration": 0.03,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.05,
+          "phone": "n_I"
+        },
+        {
+          "duration": 0.1,
+          "phone": "t_E"
+        }
+      ],
+      "start": 49.63,
+      "startOffset": 397,
+      "word": "absent"
+    },
+    {
+      "alignedWord": "from",
+      "case": "success",
+      "end": 50.46,
+      "endOffset": 408,
+      "phones": [
+        {
+          "duration": 0.06,
+          "phone": "f_B"
+        },
+        {
+          "duration": 0.04,
+          "phone": "r_I"
+        },
+        {
+          "duration": 0.06,
+          "phone": "ah_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "m_E"
+        }
+      ],
+      "start": 50.22,
+      "startOffset": 404,
+      "word": "from"
+    },
+    {
+      "alignedWord": "store",
+      "case": "success",
+      "end": 50.84,
+      "endOffset": 414,
+      "phones": [
+        {
+          "duration": 0.08,
+          "phone": "s_B"
+        },
+        {
+          "duration": 0.06,
+          "phone": "t_I"
+        },
+        {
+          "duration": 0.13,
+          "phone": "ao_I"
+        },
+        {
+          "duration": 0.08,
+          "phone": "r_E"
+        }
+      ],
+      "start": 50.49,
+      "startOffset": 409,
+      "word": "store"
+    },
+    {
+      "alignedWord": "shelves",
+      "case": "success",
+      "end": 51.42,
+      "endOffset": 422,
+      "phones": [
+        {
+          "duration": 0.2,
+          "phone": "sh_B"
+        },
+        {
+          "duration": 0.14,
+          "phone": "eh_I"
+        },
+        {
+          "duration": 0.16,
+          "phone": "l_I"
+        },
+        {
+          "duration": 0.03,
+          "phone": "v_I"
+        },
+        {
+          "duration": 0.05,
+          "phone": "z_E"
+        }
+      ],
+      "start": 50.84,
+      "startOffset": 415,
+      "word": "shelves"
+    }
+  ]
+}
\ No newline at end of file
diff --git a/tests/data/expected/full-harvard-sentences-list30-s01.json b/tests/data/expected/full-harvard-sentences-list30-s01.json
new file mode 100644
index 00000000..a57c9a06
--- /dev/null
+++ b/tests/data/expected/full-harvard-sentences-list30-s01.json
@@ -0,0 +1,223 @@
+{
+    "transcript": "the mute muscle the higher tones are who are and ",
+    "words": [
+        {
+            "alignedWord": "the",
+            "case": "success",
+            "end": 0.5900000000000001,
+            "endOffset": 3,
+            "phones": [
+                {
+                    "duration": 0.17,
+                    "phone": "dh_B"
+                },
+                {
+                    "duration": 0.08,
+                    "phone": "iy_E"
+                }
+            ],
+            "start": 0.34,
+            "startOffset": 0,
+            "word": "the"
+        },
+        {
+            "alignedWord": "mute",
+            "case": "success",
+            "end": 0.98,
+            "endOffset": 8,
+            "phones": [
+                {
+                    "duration": 0.08,
+                    "phone": "m_B"
+                },
+                {
+                    "duration": 0.11,
+                    "phone": "y_I"
+                },
+                {
+                    "duration": 0.13,
+                    "phone": "uw_I"
+                },
+                {
+                    "duration": 0.07,
+                    "phone": "t_E"
+                }
+            ],
+            "start": 0.59,
+            "startOffset": 4,
+            "word": "mute"
+        },
+        {
+            "alignedWord": "muscle",
+            "case": "success",
+            "end": 1.74,
+            "endOffset": 15,
+            "phones": [
+                {
+                    "duration": 0.23,
+                    "phone": "m_B"
+                },
+                {
+                    "duration": 0.11,
+                    "phone": "ah_I"
+                },
+                {
+                    "duration": 0.12,
+                    "phone": "s_I"
+                },
+                {
+                    "duration": 0.05,
+                    "phone": "ah_I"
+                },
+                {
+                    "duration": 0.22,
+                    "phone": "l_E"
+                }
+            ],
+            "start": 1.01,
+            "startOffset": 9,
+            "word": "muscle"
+        },
+        {
+            "alignedWord": "the",
+            "case": "success",
+            "end": 1.97,
+            "endOffset": 19,
+            "phones": [
+                {
+                    "duration": 0.11,
+                    "phone": "dh_B"
+                },
+                {
+                    "duration": 0.09,
+                    "phone": "iy_E"
+                }
+            ],
+            "start": 1.77,
+            "startOffset": 16,
+            "word": "the"
+        },
+        {
+            "alignedWord": "higher",
+            "case": "success",
+            "end": 2.29,
+            "endOffset": 26,
+            "phones": [
+                {
+                    "duration": 0.1,
+                    "phone": "hh_B"
+                },
+                {
+                    "duration": 0.18,
+                    "phone": "ay_I"
+                },
+                {
+                    "duration": 0.04,
+                    "phone": "er_E"
+                }
+            ],
+            "start": 1.97,
+            "startOffset": 20,
+            "word": "higher"
+        },
+        {
+            "alignedWord": "tones",
+            "case": "success",
+            "end": 2.86,
+            "endOffset": 32,
+            "phones": [
+                {
+                    "duration": 0.16,
+                    "phone": "t_B"
+                },
+                {
+                    "duration": 0.24,
+                    "phone": "ow_I"
+                },
+                {
+                    "duration": 0.09,
+                    "phone": "n_I"
+                },
+                {
+                    "duration": 0.08,
+                    "phone": "z_E"
+                }
+            ],
+            "start": 2.29,
+            "startOffset": 27,
+            "word": "tones"
+        },
+        {
+            "alignedWord": "are",
+            "case": "success",
+            "end": 3.12,
+            "endOffset": 36,
+            "phones": [
+                {
+                    "duration": 0.26,
+                    "phone": "er_S"
+                }
+            ],
+            "start": 2.86,
+            "startOffset": 33,
+            "word": "are"
+        },
+        {
+            "alignedWord": "who",
+            "case": "success",
+            "end": 3.43,
+            "endOffset": 40,
+            "phones": [
+                {
+                    "duration": 0.11,
+                    "phone": "hh_B"
+                },
+                {
+                    "duration": 0.2,
+                    "phone": "uw_E"
+                }
+            ],
+            "start": 3.12,
+            "startOffset": 37,
+            "word": "who"
+        },
+        {
+            "alignedWord": "are",
+            "case": "success",
+            "end": 3.49,
+            "endOffset": 44,
+            "phones": [
+                {
+                    "duration": 0.06,
+                    "phone": "er_S"
+                }
+            ],
+            "start": 3.43,
+            "startOffset": 41,
+            "word": "are"
+        },
+        {
+            "alignedWord": "and",
+            "case": "success",
+            "end": 3.64,
+            "endOffset": 48,
+            "phones": [
+                {
+                    "duration": 0.03,
+                    "phone": "ah_B"
+                },
+                {
+                    "duration": 0.05,
+                    "phone": "n_I"
+                },
+                {
+                    "duration": 0.07,
+                    "phone": "d_E"
+                }
+            ],
+            "start": 3.49,
+            "startOffset": 45,
+            "word": "and"
+        }
+    ]
+}
\ No newline at end of file
diff --git a/tests/data/input/README.md b/tests/data/input/README.md
new file mode 100644
index 00000000..1c139078
--- /dev/null
+++ b/tests/data/input/README.md
@@ -0,0 +1,5 @@
+## Data sources:
+
+Harvard Sentences audio files taken from the Open Speech Repository at http://www.voiptroubleshooter.com/open_speech/american.html
+
+The Harvard Sentences are listed at http://www.cs.columbia.edu/%7Ehgs/audio/harvard.html
diff --git a/tests/data/input/harvard-sentences-list30-s01.mp3 b/tests/data/input/harvard-sentences-list30-s01.mp3
new file mode 100644
index 00000000..6714a2cf
Binary files /dev/null and b/tests/data/input/harvard-sentences-list30-s01.mp3 differ
diff --git a/tests/data/input/harvard-sentences-list30.mp3 b/tests/data/input/harvard-sentences-list30.mp3
new file mode 100644
index 00000000..5319b966
Binary files /dev/null and b/tests/data/input/harvard-sentences-list30.mp3 differ
diff --git a/tests/data/input/harvard-sentences-list30.txt b/tests/data/input/harvard-sentences-list30.txt
new file mode 100644
index 00000000..c6f9011b
--- /dev/null
+++ b/tests/data/input/harvard-sentences-list30.txt
@@ -0,0 +1,10 @@
+The mute muffled the high tones of the horn.
+The gold ring fits only a pierced ear.
+The old pan was covered with hard fudge.
+Watch the log float in the wide river.
+The node on the stalk of wheat grew daily.
+The heap of fallen leaves was set on fire.
+Write fast, if you want to finish early.
+His shirt was clean but one button was gone.
+The barrel of beer was a brew of malt and hops.
+Tin cans are absent from store shelves.
diff --git a/tests/helpers.py b/tests/helpers.py
new file mode 100644
index 00000000..968e24a5
--- /dev/null
+++ b/tests/helpers.py
@@ -0,0 +1,23 @@
+import json
+import os
+import pytest
+
+_this_dir = os.path.dirname(os.path.realpath(__file__))
+_data_dir = os.path.join(_this_dir, "data")
+_input_dir = os.path.join(_data_dir, "input")
+_expected_dir = os.path.join(_data_dir, "expected")
+_result_dir = os.path.join(_data_dir, "tmp")
+
+def input_path(name):
+    return os.path.join(_input_dir, name)
+
+def expected_path(name):
+    return os.path.join(_expected_dir, name)
+
+def result_path(name):
+    if not os.path.isdir(_result_dir): os.makedirs(_result_dir)
+    return os.path.join(_result_dir, name)
+
+def input_data(filename):
+    with open(input_path(filename)) as fh:
+        return fh.read();
diff --git a/tests/test_forced_aligner.py b/tests/test_forced_aligner.py
new file mode 100644
index 00000000..a2d17a0f
--- /dev/null
+++ b/tests/test_forced_aligner.py
@@ -0,0 +1,38 @@
+import gentle
+import logging
+import logging.handlers
+import sys
+
+from helpers import *
+
+class TestForcedAligner:
+
+    def test_harvard30(self, tmpdir, request):
+        name = "harvard-sentences-list30"
+        transcript = input_data(name + ".txt")
+        audiofile = input_path(name + ".mp3")
+        expectedfile = expected_path("forced-" + name + ".json")
+        resultfile = result_path("forced-" + name + ".json")
+
+        expected = gentle.Transcription.from_jsonfile(expectedfile)
+        assert transcript == expected.transcript # test data consistency check
+
+        resources = gentle.Resources()
+        aligner = gentle.ForcedAligner(resources, transcript)
+
+        logger = logging.getLogger(request.node.name)
+        handler = logging.handlers.MemoryHandler(sys.maxint)
+        logger.addHandler(handler)
+        logger.setLevel('INFO')
+
+        with gentle.resampled(audiofile) as wavfile:
+            result = aligner.transcribe(wavfile, logging=logger)
+
+        log = [record.getMessage() for record in handler.buffer]
+
+        with open(resultfile, "w") as fh:
+            fh.write(result.to_json())
+
+        assert result == expected
+        assert "5 unaligned words (of 86)" in log
+        assert "after 2nd pass: 0 unaligned words (of 86)" in log
diff --git a/tests/test_full_transcriber.py b/tests/test_full_transcriber.py
new file mode 100644
index 00000000..d290eedd
--- /dev/null
+++ b/tests/test_full_transcriber.py
@@ -0,0 +1,30 @@
+import gentle
+import json
+import logging
+import logging.handlers
+import sys
+
+from helpers import *
+
+class TestFullTranscriber:
+
+    def test_harvard30_s01(self, request):
+        name = "harvard-sentences-list30-s01"
+        audiofile = input_path(name + ".mp3")
+        expectedfile = expected_path("full-" + name + ".json")
+        resultfile = result_path("full-" + name + ".json")
+
+        expected = gentle.Transcription.from_jsonfile(expectedfile)
+
+        resources = gentle.Resources()
+        transcriber = gentle.FullTranscriber(resources)
+
+        assert transcriber.available # verify language model is loaded
+
+        with gentle.resampled(audiofile) as wavfile:
+            result = transcriber.transcribe(wavfile)
+
+        with open(resultfile, "w") as fh:
+            fh.write(result.to_json())
+
+        assert result == expected
diff --git a/www/view_alignment.html b/www/view_alignment.html
index 7125c8c4..63e7cb37 100644
--- a/www/view_alignment.html
+++ b/www/view_alignment.html
@@ -258,7 +258,7 @@ <h1 class="home"><a href="/">Gentle</a></h1>
     var currentOffset = 0;
 
     wds.forEach(function(wd) {
-        if(wd['case'] == 'not-found-in-transcript') {
+        if(wd.case == 'not-found-in-transcript') {
             // TODO: show phonemes somewhere
             var txt = ' ' + wd.word;
             var $plaintext = document.createTextNode(txt);