Recognize Audio
AI-generatedSummary
Start asynchronous transcription of an audio file stored in Yandex Object Storage, submitting it for speech-to-text recognition using Yandex Cloud SpeechKit STT v3.
Inputs
- Audio URL (required) — The HTTPS URL of the audio file in Yandex Object Storage to transcribe.
- Language Code — Specifies the language code of the audio for recognition; can be set to automatic detection or a specific supported language.
- Format Specification — Choose whether to specify the audio file format using Yandex Cloud audio format codes or standard MIME types.
- Audio Format — The audio file format code when choosing to specify format by audio format; options include LPCM, OGG Opus, MP3.
- MIME Type — The MIME type of the audio when choosing to specify format by MIME type; samples include audio/wav, audio/mp3, audio/ogg, etc.
- Recognition Options — Additional optional parameters including audio channel count, sample rate (Hz), whether to use literature text normalization, profanity filtering, and speaker diarization labeling.
Output shape
a single JSON object per input item indicating transcription initiation details
The output includes success status, a unique operation ID for later result retrieval, the provided audio URL, the model used, language code, and initial operation status set to 'RUNNING'. The actual transcription text is not returned by this operation but must be fetched via a separate 'Get Recognition Results' operation using the operation ID.
Examples
Example 1: Transcribe an English audio file in WAV format with default recognition settings
Set Audio URL to the file location, Language Code to 'en-US', Format Specification to 'Audio Format', and Audio Format to 'LPCM'. Leave Recognition Options at defaults.