feat: add max edge silence parameter

airenas · Oct 30, 2024 · e57eb94 · e57eb94
1 parent 1d03a45
commit e57eb94
Show file tree

Hide file tree

Showing 4 changed files with 90 additions and 16 deletions.
diff --git a/internal/pkg/service/api/api.go b/internal/pkg/service/api/api.go
@@ -17,7 +17,8 @@ type Input struct {
 	Voice            string  `json:"voice,omitempty"`
 	Priority         int     `json:"priority,omitempty"`
 	//Possible values are: word
-	SpeechMarkTypes []string `json:"speechMarkTypes,omitempty"`
+	SpeechMarkTypes      []string `json:"speechMarkTypes,omitempty"`
+	MaxEdgeSilenceMillis *int64   `json:"maxEdgeSilenceMillis,omitempty"`
 }
 
 // SpeechMark

diff --git a/internal/pkg/service/api/internal.go b/internal/pkg/service/api/internal.go
@@ -50,18 +50,19 @@ func (e AudioFormatEnum) String() string {
 
 // TTSRequestConfig config for request
 type TTSRequestConfig struct {
-	Text             string
-	RequestID        string
-	OutputFormat     AudioFormatEnum
-	OutputTextFormat TextFormatEnum
-	OutputMetadata   []string
-	AllowCollectData bool
-	SaveTags         []string
-	Speed            float32
-	Voice            string
-	Priority         int
-	AllowedMaxLen    int
-	SSMLParts        []ssml.Part
-	AudioSuffix      string
-	SpeechMarkTypes  map[string]bool
+	Text                 string
+	RequestID            string
+	OutputFormat         AudioFormatEnum
+	OutputTextFormat     TextFormatEnum
+	OutputMetadata       []string
+	AllowCollectData     bool
+	SaveTags             []string
+	Speed                float32
+	Voice                string
+	Priority             int
+	AllowedMaxLen        int
+	SSMLParts            []ssml.Part
+	AudioSuffix          string
+	SpeechMarkTypes      map[string]bool
+	MaxEdgeSilenceMillis int64
 }
diff --git a/internal/pkg/service/configurator.go b/internal/pkg/service/configurator.go
@@ -134,11 +134,16 @@ func (c *TTSConfigutaror) Configure(r *http.Request, inText *api.Input) (*api.TT
 	if err != nil {
 		return nil, err
 	}
+	goapp.Log.Info().Str("in", goapp.Sanitize(inText.Voice)).Str("mapped", res.Voice).Msg("voice")
 	res.SpeechMarkTypes, err = getSpeechMarkTypes(inText.SpeechMarkTypes)
 	if err != nil {
 		return nil, err
 	}
-	goapp.Log.Info().Msgf("Voice '%s' -> '%s'", goapp.Sanitize(inText.Voice), res.Voice)
+	res.MaxEdgeSilenceMillis, err = getMaxEdgeSilence(inText.MaxEdgeSilenceMillis)
+	if err != nil {
+		return nil, err
+	}
+	goapp.Log.Info().Int64("edgeSil", res.MaxEdgeSilenceMillis).Any("speechMarks", res.SpeechMarkTypes).Send()
 	if inText.Priority < 0 {
 		return nil, errors.Errorf("wrong priority (>=0) value: %d", inText.Priority)
 	}
@@ -163,6 +168,16 @@ func (c *TTSConfigutaror) Configure(r *http.Request, inText *api.Input) (*api.TT
 	return res, nil
 }
 
+func getMaxEdgeSilence(value *int64) (int64, error) {
+	if value == nil {
+		return -1, nil
+	}
+	if *value < 0 {
+		return -1, errors.Errorf("maxEdgeSilenceMillis must be >= 0")
+	}
+	return *value, nil
+}
+
 func getSpeechMarkTypes(s []string) (map[string]bool, error) {
 	res := make(map[string]bool)
 	for _, v := range s {

diff --git a/internal/pkg/service/configurator_test.go b/internal/pkg/service/configurator_test.go
@@ -3,6 +3,7 @@ package service
 import (
 	"fmt"
 	"net/http/httptest"
+	"reflect"
 	"strings"
 	"testing"
 
@@ -402,3 +403,59 @@ func Test_getMaxLen(t *testing.T) {
 		})
 	}
 }
+
+func Test_getSpeechMarkTypes(t *testing.T) {
+	type args struct {
+		s []string
+	}
+	tests := []struct {
+		name    string
+		args    args
+		want    map[string]bool
+		wantErr bool
+	}{
+		{name: "OK", args: args{s: []string{"word"}}, want: map[string]bool{"word": true}, wantErr: false},
+		{name: "Fail", args: args{s: []string{"word1"}}, want: nil, wantErr: true},
+		{name: "Empty", args: args{s: nil}, want: map[string]bool{}, wantErr: false},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got, err := getSpeechMarkTypes(tt.args.s)
+			if (err != nil) != tt.wantErr {
+				t.Errorf("getSpeechMarkTypes() error = %v, wantErr %v", err, tt.wantErr)
+				return
+			}
+			if !reflect.DeepEqual(got, tt.want) {
+				t.Errorf("getSpeechMarkTypes() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}
+
+func Test_getMaxEdgeSilence(t *testing.T) {
+	type args struct {
+		value *int64
+	}
+	tests := []struct {
+		name    string
+		args    args
+		want    int64
+		wantErr bool
+	}{
+		{name: "OK", args: args{value: nil}, want: -1, wantErr: false},
+		{name: "OK", args: args{value: &[]int64{10}[0]}, want: 10, wantErr: false},
+		{name: "Fail", args: args{value: &[]int64{-10}[0]}, want: -1, wantErr: true},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got, err := getMaxEdgeSilence(tt.args.value)
+			if (err != nil) != tt.wantErr {
+				t.Errorf("getMaxEdgeSilence() error = %v, wantErr %v", err, tt.wantErr)
+				return
+			}
+			if got != tt.want {
+				t.Errorf("getMaxEdgeSilence() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}