Skip to content

fix(elevenlabs): don't persist the per-call language passed to recognize() #7822

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
Atharvsinh-codez wants to merge 1 commit into from
Closed

fix(elevenlabs): don't persist the per-call language passed to recognize() #7822

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 6 additions & 5 deletions livekit-plugins/livekit-plugins-elevenlabs/livekit/plugins/elevenlabs/stt.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -283,16 +283,17 @@ async def _recognize_impl(
language: NotGivenOr[str] = NOT_GIVEN,
conn_options: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS,
) -> stt.SpeechEvent:
if is_given(language):
self._opts.language_code = LanguageCode(language)
request_language = (
LanguageCode(language) if is_given(language) else self._opts.language_code
)

wav_bytes = rtc.combine_audio_frames(buffer).to_wav_bytes()
form = aiohttp.FormData()
form.add_field("file", wav_bytes, filename="audio.wav", content_type="audio/x-wav")
form.add_field("model_id", self._opts.model_id)
form.add_field("tag_audio_events", str(self._opts.tag_audio_events).lower())
if self._opts.language_code:
form.add_field("language_code", self._opts.language_code)
if request_language:
form.add_field("language_code", request_language)
if is_given(self._opts.keyterms):
for keyterm in self._opts.keyterms:
form.add_field("keyterms", keyterm)
Expand Down Expand Up @@ -346,7 +347,7 @@ async def _recognize_impl(
except Exception as e:
raise APIConnectionError() from e

normalized_language = LanguageCode(language_code or self._opts.language_code or "")
normalized_language = LanguageCode(language_code or request_language or "")
return self._transcription_to_speech_event(
language_code=normalized_language,
text=extracted_text,
Expand Down
26 changes: 26 additions & 0 deletions tests/test_plugin_elevenlabs_stt.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -763,3 +763,29 @@ async def handler(request: web.Request) -> web.Response:
assert exc_info.value.message == message
assert exc_info.value.body == parsed
assert requests == 1


async def test_recognize_language_override_applies_to_one_call() -> None:
# StreamAdapter passes its stream's language to every recognize() call, so an
# override must not leak into the STT's own configuration
sent_languages: list[str] = []

async def handler(request: web.Request) -> web.Response:
form = await request.post()
sent_languages.append(str(form.get("language_code")))
return web.json_response({"text": "hello", "words": []})

app = web.Application()
app.router.add_post("/speech-to-text", handler)
async with TestServer(app) as server, aiohttp.ClientSession() as session:
url = str(server.make_url(""))
batch_stt = elevenlabs_stt.STT(
api_key="test-key", base_url=url, language_code="en", http_session=session
)
overridden = await batch_stt.recognize([_frame(100)], language="fr")
default = await batch_stt.recognize([_frame(100)])

assert sent_languages == ["fr", "en"]
assert overridden.alternatives[0].language == "fr"
assert default.alternatives[0].language == "en"
assert batch_stt._opts.language_code == "en"
Loading