Merge pull request #53 from ahmetoner/add-new-large-model-v2

Add new large model v2

README.md

@@ -34,7 +34,7 @@ docker run -d -p 9000:9000 -e ASR_MODEL=base onerahmet/openai-whisper-asr-webser
 # Interactive Swagger API documentation is available at http://localhost:9000/docs
 ```
 
-Available ASR_MODELs are `tiny`, `base`, `small`, `medium` and `large`
+Available ASR_MODELs are `tiny`, `base`, `small`, `medium`, `large`, `large-v1` and `large-v2`. Please note that `large` and `large-v2` are the same model.
 
 For English-only applications, the `.en` models tend to perform better, especially for the `tiny.en` and `base.en` models. We observed that the difference becomes less significant for the `small.en` and `medium.en` models.
 
@@ -64,7 +64,7 @@ poetry install
 Starting the Webservice:
 
 ```sh
-gunicorn --bind 0.0.0.0:9000 --workers 1 --timeout 0 app.webservice:app -k uvicorn.workers.UvicornWorker
+gunicorn --bind 0.0.0.0:9001 --workers 1 --timeout 0 app.webservice:app -k uvicorn.workers.UvicornWorker
 ```
 
 ## Quick start

app/webservice.py

@@ -4,12 +4,12 @@
 from fastapi.openapi.docs import get_swagger_ui_html
 import whisper
 from whisper.utils import write_srt, write_vtt
+from whisper import tokenizer
 import os
 from os import path
 from pathlib import Path
 import ffmpeg
 from typing import BinaryIO, Union
-from .languages import LANGUAGES, LANGUAGE_CODES
 import numpy as np
 from io import StringIO
 from threading import Lock
@@ -18,6 +18,7 @@
 import importlib.metadata 
 
 SAMPLE_RATE=16000
+LANGUAGE_CODES=sorted(list(tokenizer.LANGUAGES.keys()))
 
 projectMetadata = importlib.metadata.metadata('whisper-asr-webservice')
 app = FastAPI(
@@ -101,7 +102,7 @@ def language_detection(
         _, probs = model.detect_language(mel)
     detected_lang_code = max(probs, key=probs.get)
 
-    result = { "detected_language": LANGUAGES[detected_lang_code],
+    result = { "detected_language": tokenizer.LANGUAGES[detected_lang_code],
               "langauge_code" : detected_lang_code }
 
     return result

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "whisper-asr-webservice"
-version = "1.0.4"
+version = "1.0.5"
 description = "Whisper ASR Webservice is a general-purpose speech recognition webservice."
 homepage  = "https://github.com/ahmetoner/whisper-asr-webservice/"
 license = "https://github.com/ahmetoner/whisper-asr-webservice/blob/main/LICENCE"
@@ -16,7 +16,7 @@ python = "^3.9"
 unidecode = "^1.3.4"
 uvicorn = { extras = ["standard"], version = "^0.18.2" }
 gunicorn = "^20.1.0"
-whisper = {git = "https://github.com/openai/whisper.git", rev="eff383b27b783e280c089475852ba83f20f64998"}
+whisper = {git = "https://github.com/openai/whisper.git", rev="b9265e5796f5d80c18d1f9231ab234225676780b"}
 tqdm = "^4.64.1"
 transformers = "^4.22.1"
 python-multipart = "^0.0.5"