services(cartesia): allow output_format and model_id

CHANGELOG.md

@@ -9,6 +9,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- Allow passing `output_format` and `model_id` to `CartesiaTTSService` to change
+  audio sample format and the model to use.
+
 - Added `DailyRESTHelper` which helps you create Daily rooms and tokens in an
   easy way.
 

examples/foundational/07d-interruptible-cartesia.py

@@ -39,6 +39,7 @@ async def main(room_url: str, token):
             "Respond bot",
             DailyParams(
                 audio_out_enabled=True,
+                audio_out_sample_rate=44100,
                 transcription_enabled=True,
                 vad_enabled=True,
                 vad_analyzer=SileroVADAnalyzer()
@@ -47,7 +48,8 @@ async def main(room_url: str, token):
 
         tts = CartesiaTTSService(
             api_key=os.getenv("CARTESIA_API_KEY"),
-            voice_name="Barbershop Man"
+            voice_name="British Lady",
+            output_format="pcm_44100"
         )
 
         llm = OpenAILLMService(

src/pipecat/services/cartesia.py

@@ -21,11 +21,15 @@ class CartesiaTTSService(TTSService):
             *,
             api_key: str,
             voice_name: str,
+            model_id: str = "upbeat-moon",
+            output_format: str = "pcm_16000",
             **kwargs):
         super().__init__(**kwargs)
 
         self._api_key = api_key
         self._voice_name = voice_name
+        self._model_id = model_id
+        self._output_format = output_format
 
         try:
             self._client = AsyncCartesiaTTS(api_key=self._api_key)
@@ -40,14 +44,14 @@ class CartesiaTTSService(TTSService):
 
         try:
             chunk_generator = await self._client.generate(
-                transcript=text, voice=self._voice, stream=True,
+                stream=True,
-                model_id="upbeat-moon", data_rtype='array', output_format='pcm_16000',
+                transcript=text,
-                # a chunk_time of 0.1 seems to be the default. there are small audio pops/gaps which
+                voice=self._voice,
-                # we need to debug
+                model_id=self._model_id,
-                chunk_time=0.1
+                output_format=self._output_format,
             )
 
             async for chunk in chunk_generator:
-                yield AudioRawFrame(chunk['audio'], 16000, 1)
+                yield AudioRawFrame(chunk["audio"], chunk["sampling_rate"], 1)
         except Exception as e:
             logger.error(f"Cartesia exception: {e}")