Class GeminiAudioTranscriptionCapabilityImpl

Description

Provider capability implementation for GeminiAudioTranscriptionCapabilityImpl.

Implements

Index

Constructors

Methods

transcribeAudio transcribeAudioStream

Constructors

constructor

new GeminiAudioTranscriptionCapabilityImpl(
provider: BaseProvider,
client: GoogleGenAI,
): GeminiAudioTranscriptionCapabilityImpl
Parameters
- provider: BaseProvider
- client: GoogleGenAI
Returns GeminiAudioTranscriptionCapabilityImpl
- Defined in src/providers/gemini/capabilities/GeminiAudioTranscriptionCapabilityImpl.ts:38

Methods

transcribeAudio

transcribeAudio(
    request: AIRequest<ClientAudioTranscriptionRequest>,
    _ctx: MultiModalExecutionContext,
    signal?: AbortSignal,
): Promise<AIResponse<NormalizedChatMessage[]>>
Transcribes audio in non-streaming mode.

Steps:
- Validate input
- Merge provider/model options
- Normalize audio payload
- Build Gemini instruction prompt
- Call generateContent and normalize output
Parameters
- request: AIRequest<ClientAudioTranscriptionRequest>
  Unified AI request containing transcription input/options/context
- _ctx: MultiModalExecutionContext
  Optional multimodal execution context (unused by this capability)
- Optionalsignal: AbortSignal
  Optional abort signal
Returns Promise<AIResponse<NormalizedChatMessage[]>>
Provider-normalized transcript messages

Throws
If input is invalid or request is aborted
Implementation of AudioTranscriptionCapability.transcribeAudio
- Defined in src/providers/gemini/capabilities/GeminiAudioTranscriptionCapabilityImpl.ts:60

transcribeAudioStream

transcribeAudioStream(
    request: AIRequest<ClientAudioTranscriptionRequest>,
    _ctx: MultiModalExecutionContext,
    signal?: AbortSignal,
): AsyncGenerator<AIResponseChunk<NormalizedChatMessage[]>>
Transcribes audio in streaming mode and emits incremental transcript deltas.

Emits:
- done: false chunks while transcript text accumulates
- a terminal done: true completed chunk
- a terminal done: true error chunk on failure
Parameters
- request: AIRequest<ClientAudioTranscriptionRequest>
  Unified AI request containing transcription input/options/context
- _ctx: MultiModalExecutionContext
  Optional multimodal execution context (unused by this capability)
- Optionalsignal: AbortSignal
  Optional abort signal
Returns AsyncGenerator<AIResponseChunk<NormalizedChatMessage[]>>
Async generator of delta and final transcript chunks

Throws
If input is invalid before streaming starts
Implementation of AudioTranscriptionStreamCapability.transcribeAudioStream
- Defined in src/providers/gemini/capabilities/GeminiAudioTranscriptionCapabilityImpl.ts:158