wx44wx/pipecat
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

NvidiaSTTService: initialize client on StartFrame

Browse Source 
Initialize client on StartFrame so errrors are reported within the pipeline.
This commit is contained in:
Aleix Conchillo Flaqué
2026-01-19 20:32:28 -08:00
parent 9a718ded1e
commit 671dc8cd9b
1 changed files with 49 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

91
src/pipecat/services/nvidia/stt.py
View File

@@ -134,6 +134,7 @@ class NvidiaSTTService(STTService):
params = params or NvidiaSTTService.InputParams()
self._server = server
self._api_key = api_key
self._use_ssl = use_ssl
self._profanity_filter = False
@@ -162,19 +163,55 @@ class NvidiaSTTService(STTService):
self.set_model_name(model_function_map.get("model_name"))
metadata = [
["function-id", self._function_id],
["authorization", f"Bearer {api_key}"],
]
auth = riva.client.Auth(None, self._use_ssl, server, metadata)
self._asr_service = riva.client.ASRService(auth)
self._asr_service = None
self._queue = None
self._config = None
self._thread_task = None
self._response_task = None
def _initialize_client(self):
metadata = [
["function-id", self._function_id],
["authorization", f"Bearer {self._api_key}"],
]
auth = riva.client.Auth(None, self._use_ssl, self._server, metadata)
self._asr_service = riva.client.ASRService(auth)
def _create_recognition_config(self):
"""Create the NVIDIA Riva ASR recognition configuration."""
config = riva.client.StreamingRecognitionConfig(
config=riva.client.RecognitionConfig(
encoding=riva.client.AudioEncoding.LINEAR_PCM,
language_code=self._language_code,
model="",
max_alternatives=1,
profanity_filter=self._profanity_filter,
enable_automatic_punctuation=self._automatic_punctuation,
verbatim_transcripts=not self._no_verbatim_transcripts,
sample_rate_hertz=self.sample_rate,
audio_channel_count=1,
),
interim_results=True,
)
riva.client.add_word_boosting_to_config(
config, self._boosted_lm_words, self._boosted_lm_score
)
riva.client.add_endpoint_parameters_to_config(
config,
self._start_history,
self._start_threshold,
self._stop_history,
self._stop_history_eou,
self._stop_threshold,
self._stop_threshold_eou,
)
riva.client.add_custom_configuration_to_config(config, self._custom_configuration)
return config
def can_generate_metrics(self) -> bool:
"""Check if this service can generate processing metrics.
@@ -206,41 +243,9 @@ class NvidiaSTTService(STTService):
frame: StartFrame indicating pipeline start.
"""
await super().start(frame)
self._initialize_client()
self._config = self._create_recognition_config()
if self._config:
return
config = riva.client.StreamingRecognitionConfig(
config=riva.client.RecognitionConfig(
encoding=riva.client.AudioEncoding.LINEAR_PCM,
language_code=self._language_code,
model="",
max_alternatives=1,
profanity_filter=self._profanity_filter,
enable_automatic_punctuation=self._automatic_punctuation,
verbatim_transcripts=not self._no_verbatim_transcripts,
sample_rate_hertz=self.sample_rate,
audio_channel_count=1,
),
interim_results=True,
)
riva.client.add_word_boosting_to_config(
config, self._boosted_lm_words, self._boosted_lm_score
)
riva.client.add_endpoint_parameters_to_config(
config,
self._start_history,
self._start_threshold,
self._stop_history,
self._stop_history_eou,
self._stop_threshold,
self._stop_threshold_eou,
)
riva.client.add_custom_configuration_to_config(config, self._custom_configuration)
self._config = config
self._queue = asyncio.Queue()
if not self._thread_task:
@@ -250,6 +255,8 @@ class NvidiaSTTService(STTService):
self._response_queue = asyncio.Queue()
self._response_task = self.create_task(self._response_task_handler())
logger.debug(f"Initialized NvidiaSTTService with model: {self.model_name}")
async def stop(self, frame: EndFrame):
"""Stop the NVIDIA Riva STT service and clean up resources.