mirror of
https://github.com/dograh-hq/dograh.git
synced 2026-07-22 11:51:04 +02:00
feat(tts): add websocket transport option for xAI TTS (#548)
* feat(tts): add websocket transport option for xAI TTS
xAI's pipecat service ships two implementations: XAIHttpTTSService
(batch REST, used by the existing xAI integration) and XAITTSService
(realtime WebSocket streaming). This adds a transport field on
XAITTSConfiguration ("http", default, unchanged behavior | "websocket")
and wires the latter into create_tts_service(), so live voice calls can
opt into lower time-to-first-byte streaming instead of request/response.
Originally built and battle-tested independently against v1.41.0 in a
production deployment (7 live Telnyx calls, 2026-07-15) before this PR
existed; ported onto current main and given a config toggle instead of
a second provider entry so it composes with the existing xAI HTTP path
rather than duplicating it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
* refactor(tts): use websocket-only xAI TTS, drop the transport toggle
---------
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
Co-authored-by: Sabiha Khan <sabihak89@gmail.com>
This commit is contained in:
parent
177bbe6885
commit
94dce42209
4 changed files with 13 additions and 27 deletions
|
|
@ -1332,7 +1332,6 @@ class XAITTSConfiguration(BaseServiceConfiguration):
|
|||
description="BCP-47 language code for synthesis (e.g. 'en', 'fr', 'de'), or 'auto' for automatic language detection.",
|
||||
json_schema_extra={"allow_custom_input": True},
|
||||
)
|
||||
|
||||
@computed_field
|
||||
@property
|
||||
def model(self) -> str:
|
||||
|
|
|
|||
|
|
@ -86,7 +86,7 @@ from pipecat.services.speechmatics.stt import (
|
|||
SpeechmaticsSTTService,
|
||||
SpeechmaticsSTTSettings,
|
||||
)
|
||||
from pipecat.services.xai.tts import XAIHttpTTSService, XAITTSSettings
|
||||
from pipecat.services.xai.tts import XAITTSService, XAIWebsocketTTSSettings
|
||||
from pipecat.transcriptions.language import Language
|
||||
from pipecat.utils.text.xml_function_tag_filter import XMLFunctionTagFilter
|
||||
|
||||
|
|
@ -817,11 +817,9 @@ def create_tts_service(
|
|||
pipecat_language = Language(language_code)
|
||||
except ValueError:
|
||||
pipecat_language = Language.EN
|
||||
return XAIHttpTTSService(
|
||||
return XAITTSService(
|
||||
api_key=user_config.tts.api_key,
|
||||
sample_rate=audio_config.transport_out_sample_rate,
|
||||
encoding="pcm",
|
||||
settings=XAITTSSettings(
|
||||
settings=XAIWebsocketTTSSettings(
|
||||
voice=voice,
|
||||
language=pipecat_language,
|
||||
),
|
||||
|
|
|
|||
|
|
@ -24,10 +24,7 @@ def test_xai_tts_configuration_defaults():
|
|||
assert XAI_TTS_VOICES == ["eve", "ara", "leo", "rex", "sal"]
|
||||
|
||||
|
||||
@pytest.mark.parametrize("transport_out_sample_rate", [8000, 16000])
|
||||
def test_create_xai_tts_service_uses_pipeline_compatible_audio_format(
|
||||
transport_out_sample_rate,
|
||||
):
|
||||
def test_create_xai_tts_service_uses_websocket_streaming():
|
||||
user_config = SimpleNamespace(
|
||||
tts=SimpleNamespace(
|
||||
provider=ServiceProviders.XAI.value,
|
||||
|
|
@ -38,20 +35,18 @@ def test_create_xai_tts_service_uses_pipeline_compatible_audio_format(
|
|||
)
|
||||
)
|
||||
audio_config = SimpleNamespace(
|
||||
transport_out_sample_rate=transport_out_sample_rate,
|
||||
transport_out_sample_rate=24000,
|
||||
transport_in_sample_rate=16000,
|
||||
)
|
||||
|
||||
with patch(
|
||||
"api.services.pipecat.service_factory.XAIHttpTTSService"
|
||||
) as mock_service:
|
||||
with patch("api.services.pipecat.service_factory.XAITTSService") as mock_service:
|
||||
create_tts_service(user_config, audio_config)
|
||||
|
||||
assert mock_service.call_count == 1
|
||||
kwargs = mock_service.call_args.kwargs
|
||||
assert kwargs["api_key"] == "test-key"
|
||||
assert kwargs["sample_rate"] == transport_out_sample_rate
|
||||
assert kwargs["encoding"] == "pcm"
|
||||
# Sample rate is resolved from the pipeline StartFrame, like other providers.
|
||||
assert "sample_rate" not in kwargs
|
||||
assert kwargs["settings"].voice == "rex"
|
||||
assert kwargs["settings"].language == Language.EN
|
||||
|
||||
|
|
@ -71,9 +66,7 @@ def test_create_xai_tts_service_converts_language():
|
|||
transport_in_sample_rate=16000,
|
||||
)
|
||||
|
||||
with patch(
|
||||
"api.services.pipecat.service_factory.XAIHttpTTSService"
|
||||
) as mock_service:
|
||||
with patch("api.services.pipecat.service_factory.XAITTSService") as mock_service:
|
||||
create_tts_service(user_config, audio_config)
|
||||
|
||||
kwargs = mock_service.call_args.kwargs
|
||||
|
|
@ -95,9 +88,7 @@ def test_create_xai_tts_service_falls_back_to_english_for_unknown_language():
|
|||
transport_in_sample_rate=16000,
|
||||
)
|
||||
|
||||
with patch(
|
||||
"api.services.pipecat.service_factory.XAIHttpTTSService"
|
||||
) as mock_service:
|
||||
with patch("api.services.pipecat.service_factory.XAITTSService") as mock_service:
|
||||
create_tts_service(user_config, audio_config)
|
||||
|
||||
kwargs = mock_service.call_args.kwargs
|
||||
|
|
@ -119,9 +110,7 @@ def test_create_xai_tts_service_preserves_auto_language():
|
|||
transport_in_sample_rate=16000,
|
||||
)
|
||||
|
||||
with patch(
|
||||
"api.services.pipecat.service_factory.XAIHttpTTSService"
|
||||
) as mock_service:
|
||||
with patch("api.services.pipecat.service_factory.XAITTSService") as mock_service:
|
||||
create_tts_service(user_config, audio_config)
|
||||
|
||||
kwargs = mock_service.call_args.kwargs
|
||||
|
|
|
|||
4
ui/package-lock.json
generated
4
ui/package-lock.json
generated
|
|
@ -1,12 +1,12 @@
|
|||
{
|
||||
"name": "ui",
|
||||
"version": "1.41.0",
|
||||
"version": "1.42.0",
|
||||
"lockfileVersion": 3,
|
||||
"requires": true,
|
||||
"packages": {
|
||||
"": {
|
||||
"name": "ui",
|
||||
"version": "1.41.0",
|
||||
"version": "1.42.0",
|
||||
"dependencies": {
|
||||
"@calcom/embed-react": "^1.5.3",
|
||||
"@dagrejs/dagre": "^1.1.4",
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue