Merge pull request #316 from JCallicoat/volume_setting

Add a volume multiplier setting

api/src/core/config.py

@@ -31,6 +31,7 @@ class Settings(BaseSettings):
 
     # Audio Settings
     sample_rate: int = 24000
+    default_volume_multiplier: float = 1.0
     # Text Processing Settings
     target_min_tokens: int = 175  # Target minimum tokens per chunk
     target_max_tokens: int = 250  # Target maximum tokens per chunk

api/src/inference/model_manager.py

@@ -141,6 +141,8 @@ Model files not found! You need to download the Kokoro V1 model:
 
         try:
             async for chunk in self._backend.generate(*args, **kwargs):
+                if settings.default_volume_multiplier != 1.0:
+                    chunk.audio *= settings.default_volume_multiplier
                 yield chunk
         except Exception as e:
             raise RuntimeError(f"Generation failed: {e}")

api/src/routers/development.py

@@ -319,6 +319,7 @@ async def create_captioned_speech(
                 writer=writer,
                 speed=request.speed,
                 return_timestamps=request.return_timestamps,
+                volume_multiplier=request.volume_multiplier,
                 normalization_options=request.normalization_options,
                 lang_code=request.lang_code,
             )

api/src/routers/openai_compatible.py

@@ -152,6 +152,7 @@ async def stream_audio_chunks(
             speed=request.speed,
             output_format=request.response_format,
             lang_code=request.lang_code,
+            volume_multiplier=request.volume_multiplier,
             normalization_options=request.normalization_options,
             return_timestamps=unique_properties["return_timestamps"],
         ):
@@ -300,6 +301,7 @@ async def create_speech(
                 voice=voice_name,
                 writer=writer,
                 speed=request.speed,
+                volume_multiplier=request.volume_multiplier,
                 normalization_options=request.normalization_options,
                 lang_code=request.lang_code,
             )

api/src/services/tts_service.py

@@ -55,6 +55,7 @@ class TTSService:
         output_format: Optional[str] = None,
         is_first: bool = False,
         is_last: bool = False,
+        volume_multiplier: Optional[float] = 1.0,
         normalizer: Optional[AudioNormalizer] = None,
         lang_code: Optional[str] = None,
         return_timestamps: Optional[bool] = False,
@@ -100,6 +101,7 @@ class TTSService:
                         lang_code=lang_code,
                         return_timestamps=return_timestamps,
                     ):
+                        chunk_data.audio*=volume_multiplier
                         # For streaming, convert to bytes
                         if output_format:
                             try:
@@ -132,7 +134,7 @@ class TTSService:
                         speed=speed,
                         return_timestamps=return_timestamps,
                     )
-
+                    
                     if chunk_data.audio is None:
                         logger.error("Model generated None for audio chunk")
                         return
@@ -141,6 +143,8 @@ class TTSService:
                         logger.error("Model generated empty audio chunk")
                         return
 
+                    chunk_data.audio*=volume_multiplier
+
                     # For streaming, convert to bytes
                     if output_format:
                         try:
@@ -259,6 +263,7 @@ class TTSService:
         speed: float = 1.0,
         output_format: str = "wav",
         lang_code: Optional[str] = None,
+        volume_multiplier: Optional[float] = 1.0,
         normalization_options: Optional[NormalizationOptions] = NormalizationOptions(),
         return_timestamps: Optional[bool] = False,
     ) -> AsyncGenerator[AudioChunk, None]:
@@ -300,6 +305,7 @@ class TTSService:
                             formatted_pause_chunk = await AudioService.convert_audio(
                                 pause_chunk, output_format, writer, speed=speed, chunk_text="",
                                 is_last_chunk=False, trim_audio=False, normalizer=stream_normalizer,
+
                             )
                             if formatted_pause_chunk.output:
                                 yield formatted_pause_chunk
@@ -330,6 +336,7 @@ class TTSService:
                             writer,
                             output_format,
                             is_first=(chunk_index == 0),
+                            volume_multiplier=volume_multiplier,
                             is_last=False,  # We'll update the last chunk later
                             normalizer=stream_normalizer,
                             lang_code=pipeline_lang_code,  # Pass lang_code
@@ -377,6 +384,7 @@ class TTSService:
                         output_format,
                         is_first=False,
                         is_last=True,  # Signal this is the last chunk
+                        volume_multiplier=volume_multiplier,
                         normalizer=stream_normalizer,
                         lang_code=pipeline_lang_code,  # Pass lang_code
                     ):
@@ -396,6 +404,7 @@ class TTSService:
         writer: StreamingAudioWriter,
         speed: float = 1.0,
         return_timestamps: bool = False,
+        volume_multiplier: Optional[float] = 1.0,
         normalization_options: Optional[NormalizationOptions] = NormalizationOptions(),
         lang_code: Optional[str] = None,
     ) -> AudioChunk:
@@ -408,6 +417,7 @@ class TTSService:
                 voice,
                 writer,
                 speed=speed,
+                volume_multiplier=volume_multiplier,
                 normalization_options=normalization_options,
                 return_timestamps=return_timestamps,
                 lang_code=lang_code,

api/src/structures/schemas.py

@@ -113,6 +113,10 @@ class OpenAISpeechRequest(BaseModel):
         default=None,
         description="Optional language code to use for text processing. If not provided, will use first letter of voice name.",
     )
+    volume_multiplier: Optional[float] = Field(
+        default = 1.0,
+        description="A volume multiplier to multiply the output audio by."
+    )
     normalization_options: Optional[NormalizationOptions] = Field(
         default=NormalizationOptions(),
         description="Options for the normalization system",
@@ -157,6 +161,10 @@ class CaptionedSpeechRequest(BaseModel):
         default=None,
         description="Optional language code to use for text processing. If not provided, will use first letter of voice name.",
     )
+    volume_multiplier: Optional[float] = Field(
+        default = 1.0,
+        description="A volume multiplier to multiply the output audio by."
+    )
     normalization_options: Optional[NormalizationOptions] = Field(
         default=NormalizationOptions(),
         description="Options for the normalization system",