Speech-to-Text

#Use it when

• You want voice notes or call transcripts
• You need captions or searchable media

#Code examples

1curl -X POST https://api.navy/v1/audio/transcriptions \
2  -H "Authorization: Bearer sk-navy-YOUR_KEY" \
3  -F file="@/path/to/audio.mp3" \
4  -F model="whisper-1"

#Parameters

• model (string, required) — whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe, scribe_v2
• file (file, required) — Audio file (max 25MB). Common formats: mp3, mp4, mpeg, mpga, m4a, wav, webm.
• language (string, optional) — ISO-639-1 language code (e.g. "en", "de", "ja"). Improves accuracy and speed.
• prompt (string, optional) — Optional hint string containing names, jargon, or expected phrasing to bias transcription.
• temperature (number, optional) — 0.0–1.0. Higher values produce more variation.
• response_format (string, optional) — json (default), text, srt, vtt, verbose_json. Use verbose_json to get segments and metadata.
• timestamp_granularities (array, optional) — Requires verbose_json. Accepts ["word"], ["segment"], or both. Adds per-word or per-segment timestamps to the response.

#Notes

• All fields are sent as multipart form data, including model and any optional parameters.
• Use verbose_json + timestamp_granularities for caption generation and audio search.
• Pair this with embeddings or chat completions for analysis workflows.