Azure Speech

Fast Transcription for word and phrase timestamps and diarization

azure-speech-fast-transcription is a supported speech-to-text option for transcripts that need precise timing or speaker information. It is the only active Azure Speech route. Standard speech-to-text requests may use supported OpenAI transcription models instead.

Upload media to private Gencow storage, then pass its storageId to the generated AI facade:

import { ai } from "./ai";

const job = await ai.speech.transcribe({
    storageId: uploadedAudio.storageId,
    model: "azure-speech-fast-transcription",
    language: "ko",
    idempotencyKey: requestId,
});

const result = await ai.speech.waitForTranscript(job.id);

The platform handles the provider credential, private result storage, and service-credit billing. The base meter is 1 credit per audio second before the plan's service markup. The live plan policy controls the final charge.

Azure Speech text-to-speech and other Azure AI routes are unavailable through the hosted gateway. For text-to-speech, use OpenAI gpt-4o-mini-tts through ai.speech.synthesize() or an async synthesis job. See AI Engine for the current model and billing rules.

RAG & MemoryCLI Reference