allow tts param definition in config file

- consolidate TTS options under `tts` config entry

Signed-off-by: blob42 <[email protected]>

Author: blob42

core/config/backend_config.go

@@ -15,6 +15,15 @@ const (
 	RAND_SEED = -1
 )
 
+type TTSConfig struct {
+
+	// Voice wav path or id
+	Voice string `yaml:"voice"`
+
+	// Vall-e-x
+	VallE    VallE  `yaml:"vall-e"`
+}
+
 type BackendConfig struct {
 	schema.PredictionOptions `yaml:"parameters"`
 	Name                     string `yaml:"name"`
@@ -49,8 +58,8 @@ type BackendConfig struct {
 	// GRPC Options
 	GRPC GRPC `yaml:"grpc"`
 
-	// Vall-e-x
-	VallE VallE `yaml:"vall-e"`
+	// TTS specifics
+	TTSConfig `yaml:"tts"`
 
 	// CUDA
 	// Explicitly enable CUDA or not (some backends might need it)

core/http/endpoints/localai/tts.go

@@ -43,6 +43,7 @@ func TTSEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, appConfi
 		)
 
 		if err != nil {
+			log.Err(err)
 			modelFile = input.Model
 			log.Warn().Msgf("Model not found in context: %s", input.Model)
 		} else {
@@ -54,7 +55,15 @@ func TTSEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, appConfi
 			cfg.Backend = input.Backend
 		}
 
-		filePath, _, err := backend.ModelTTS(cfg.Backend, input.Input, modelFile, input.Voice, input.Language, ml, appConfig, *cfg)
+		if input.Language != "" {
+			cfg.Language = input.Language
+		}
+
+		if input.Voice != "" {
+			cfg.Voice = input.Voice
+		}
+
+		filePath, _, err := backend.ModelTTS(cfg.Backend, input.Input, modelFile, cfg.Voice, cfg.Language, ml, appConfig, *cfg)
 		if err != nil {
 			return err
 		}