Transcription models available - Question on gpt-4o-transcribe gpt-4o-mini-transcribe gpt-4o-transcribe-diarize

@vb

About this “gpt-4o-transcribe-diarize” GPT-4o Transcribe Diarize Model | OpenAI API

The problem is that the API doesn’t allow setting response format to “ddiarized_json” which is needed for this model. There’s an open issue about this here https://github.com/openai/openai-dotnet/issues/869

Seems like the required “chunking_strategy” is not currently supported by the SDK and we get HTTP 400 (invalid_request_error: invalid_value) Parameter: chunking_strategy chunking_strategy is required for diarization models

This shows that there is active work ongoing.

github.com/openai/openai-dotnet

[Audio] Add support for diarized audio transcription (#1008#event-23660087929)

main ← ShivangiReja:Shreja_DiarizedAudioTranscription

opened 11:38PM - 05 Mar 26 UTC

        ShivangiReja
      

+3171 -622

Summary Adds client support for the diarized audio transcription. ## Ch…anges ### New API surface - AudioClient.TranscribeAudioDiarized / TranscribeAudioDiarizedAsync — four new method overloads (sync/async × Stream/file path) that return DiarizedAudioTranscription - DiarizedAudioTranscription — response model containing Text, Duration, Segments, and Usage - DiarizedTranscriptionSegment — segment model with Speaker, Text, StartTime, EndTime, and Logprob - TranscriptionUsage — base usage type with discriminated subtypes: - TranscriptionTokenUsage (with InputTokens, OutputTokens, TotalTokens, InputTokenDetails) - TranscriptionDurationUsage (with Seconds) - TranscriptionUsageKind — extensible enum (Tokens, Duration) - TranscriptionInputTokenDetails — TextTokens, AudioTokens - AudioTranscriptionFormat.DiarizedJson — now public - AudioTranscriptionOptions.KnownSpeakerNames and KnownSpeakerReferences — now public ### Tests & examples - Five new tests: DiarizedTranscriptionWorks, DiarizedTranscriptionWithKnownSpeakersWorks, DiarizedTranscriptionHasUsage, DiarizedTranscriptionSegmentsAreOrdered, TranscriptionUsageWorks - Two new examples: Example05_DiarizedTranscription (sync and async) - New test model entry: gpt-4o-transcribe-diarize Fixes: https://github.com/openai/openai-dotnet/issues/916

Guess I will wait and watch… If you happen to have any additional news, do not hesitate and let me know.

Thanks so much Dears