Module transcriptions 

Transcribes audio into the input language.

Endpoint: POST /audio/transcriptions (multipart/form-data request).

Response shapes depend on response_format:

• json and verbose_json deserialize into the typed TranscriptionResponse via get_response.
• text, srt, and vtt return plain text; use get_response_string for those.

Streaming transcriptions (stream) is not supported yet.

![warn] This module is untested! No OpenAI-compatible provider accessible to this project implements this endpoint, and no OpenAI API key was available for testing. If you encounter any issues, please report them on the repository.

Structs

ServerVadConfig

Manual VAD detection parameters, sent as the server_vad chunking strategy.

Transcription

Represents a transcription response returned by the model.

TranscriptionLanguage

A language detected in transcribed audio.

TranscriptionLogprob

The log probability of a token in the transcription.

TranscriptionRequest

Transcribes audio into the input language.

TranscriptionSegment

A segment of the transcribed text and its corresponding details.

TranscriptionVerbose

Represents a verbose json transcription response.

TranscriptionVerboseUsage

Usage statistics for models billed by audio input duration.

TranscriptionWord

An extracted word and its corresponding timestamp.

UsageTokensInputTokenDetails

Details about the input tokens billed for a request.

Enums

ChunkingStrategy

Controls how the audio is cut into chunks.

Include

Additional information to include in the transcription response.

TimestampGranularity

The timestamp granularities to populate for a transcription.

TranscriptionResponse

The typed transcription response: json yields a plain Transcription, verbose_json yields a TranscriptionVerbose.

TranscriptionUsage

Usage statistics for a transcription request. Billed either by token usage or by audio input duration, discriminated by type.