langgenius/dify · error · ProviderNotSupportSpeechToTextError

provider_not_support_speech_to_text

provider_not_support_speech_to_text

Error message

Provider not support speech to text.

What it means

HTTP 400 with error_code `provider_not_support_speech_to_text`, raised by ProviderNotSupportSpeechToTextError when AudioService raises ProviderNotSupportSpeechToTextServiceError. The configured model provider does not offer a speech-to-text model, so transcription cannot run. This is a capability mismatch, not a credential or quota problem.

Source

Thrown at api/controllers/console/explore/trial.py:690

            response = AudioService.transcript_asr(
                app_model=app_model,
                file=file,
                session=db.session(),
                end_user=None,
            )
            RecommendedAppService.add_trial_app_record(app_id, user_id, session=db.session())
            return response
        except services.errors.app_model_config.AppModelConfigBrokenError:
            logger.exception("App model config broken.")
            raise AppUnavailableError()
        except NoAudioUploadedServiceError:
            raise NoAudioUploadedError()
        except AudioTooLargeServiceError as e:
            raise AudioTooLargeError(str(e))
        except UnsupportedAudioTypeServiceError:
            raise UnsupportedAudioTypeError()
        except ProviderNotSupportSpeechToTextServiceError:
            raise ProviderNotSupportSpeechToTextError()
        except SpeechToTextDisabledServiceError:
            raise SpeechToTextDisabledError()
        except ProviderTokenNotInitError as ex:
            raise ProviderNotInitializeError(ex.description)
        except QuotaExceededError:
            raise ProviderQuotaExceededError()
        except ModelCurrentlyNotSupportError:
            raise ProviderModelCurrentlyNotSupportError()
        except InvokeError as e:
            raise CompletionRequestError(e.description)
        except ValueError as e:
            raise e
        except Exception as e:
            logger.exception("internal server error.")
            raise InternalServerError()


class TrialChatTextApi(TrialAppResource):

View on GitHub (pinned to ef8544b173)

Solutions

1. Configure a provider that supports speech-to-text (e.g. OpenAI Whisper) in Settings -> Model Provider and select it as the STT model.
2. In the app studio, point the speech-to-text setting at a model/provider that has the STT capability.
3. If no STT-capable provider is available, disable the audio input in the app rather than exposing a broken control.

Example fix

// before - app STT bound to a provider without STT
speech_to_text:
  enabled: true
  provider: anthropic  // no STT model

// after - bind to an STT-capable provider
speech_to_text:
  enabled: true
  provider: openai
  model: whisper-1

const providers = await fetch('/console/api/workspaces/current/model-providers').then(r => r.json())
const hasSTT = (providers.data || []).some(p => p.supported_models?.speech_to_text?.length > 0 && p.credentials_validated)
if (!hasSTT) throw new Error('configure an STT-capable provider')

try {
  const r = await fetch(audioUrl, { method: 'POST', body: fd })
  if (r.status === 400) {
    const body = await r.json()
    if (body.code === 'provider_not_support_speech_to_text') promptConfigureSTT()
  }
} catch (e) { reportToUser(e) }

When it happens

Trigger: POST TrialChatAudioApi where the tenant's active provider (or the hosted trial provider) has no STT model configured/available. Credentials are valid but the provider simply lacks a speech-to-text capability.

Common situations: Tenant configured an LLM-only provider (e.g. an Anthropic-only setup) and tried audio; the provider plugin was installed without its STT model registered; using a provider whose STT was deprecated.

Related errors

• speech_to_text_disabled
• no_audio_uploaded
• audio_too_large
• unsupported_audio_type
• Message not found

AI-assisted analysis of langgenius/dify@ef8544b173 (2026-08-12). Data as JSON: /api/errors/b267ac3312592733. Report an issue: GitHub.