> ## Documentation Index > Fetch the complete documentation index at: https://developers.mihu.ai/llms.txt > Use this file to discover all available pages before exploring further. # Transcribe a file > Native or OpenAI-compatible by `model`. With `webhook_url` the job runs asynchronously and answers `202`. For OpenAI model names see [OpenAI-compatible API](/stt/openai). ## OpenAPI ````yaml /stt/openapi.json post /v1/audio/transcriptions openapi: 3.0.3 info: title: Umay STT API version: 1.0.25 description: >- Speech recognition: files, voiceprints and audio isolation. Live audio uses the WebSocket endpoints in the guide pages. servers: - url: https://stt.eu.mihu.ai description: Europe (EU) - url: https://stt.tr.mihu.ai description: Türkiye (TR) security: - bearerAuth: [] tags: - name: Service - name: Transcription - name: Voiceprints - name: Audio isolation paths: /v1/audio/transcriptions: post: tags: - Transcription summary: Transcribe a file description: >- Native or OpenAI-compatible by `model`. With `webhook_url` the job runs asynchronously and answers `202`. For OpenAI model names see [OpenAI-compatible API](/stt/openai). operationId: createTranscription requestBody: required: true content: multipart/form-data: schema: type: object properties: file: type: string description: The audio. Any common format (WAV, MP3, ...). format: binary model: type: string description: >- `umay` (most accurate; files), or an OpenAI name such as `whisper-1`, `gpt-4o-transcribe`, `gpt-4o-transcribe-diarize`. example: umay language: type: string description: Language of the speech. enum: - tr - en - de - fr - it - ru - es example: tr normalize: type: boolean description: >- Inverse text normalisation on the final text: "bin iki yüz elli lira" becomes "1.250 TL". `false` returns the spoken words. default: true dialect_correction: type: boolean description: >- Turkish, with `normalize=true`: regional and colloquial forms in standard Turkish. default: true place_correction: type: boolean description: >- Turkish, with `normalize=true`: misheard or split place names corrected. default: true abbreviations: type: boolean description: >- Turkish, with `normalize=true`: `false` writes units out ("43 dakika"), `true` abbreviates ("43 dk"). default: false mode: type: string description: >- `normal`: files with `model=umay` use the language model where the model has one. `fast`: never. enum: - fast - normal default: fast hint_context: type: string description: >- Built-in hint lists (Turkish): `mugla`, `izmir`, `manisa`, `aydin`, `denizli`; comma-separated for several. hints: type: string description: >- Own phrases (names, places) to favour, comma or newline separated; `phrase :3` gives one phrase its own boost. hints_score: type: number description: Boost for phrases without their own. default: 2 gender_detection: type: boolean description: >- Gender of the speaker(s) with a confidence `percent` (50-100). default: false diarize: type: boolean description: Speaker label per segment (`S1`, `S2`, ...). default: false max_speakers: type: integer description: Upper bound for diarization. default: 8 speaker_lock: type: boolean description: >- Only the first speaker's speech becomes text; other voices come as `ignored`. default: false lock_threshold: type: number description: >- With `speaker_lock`: the similarity a segment needs to count as `S1` (0-1). Default 0.5, or 0.55 with `reference_voiceprint`. voice_isolation: type: boolean description: >- Removes other voices before recognition (tr, de, fr, it, es, ru); switches on only after another voice is heard. default: false reference_voiceprint: type: string description: >- Id or token of a stored voiceprint: with `voice_isolation`, `speaker_lock` or `diarize`, speaker `S1` is that person from the first second. response_format: type: string description: With OpenAI model names. enum: - json - text - srt - vtt - verbose_json - diarized_json default: json webhook_url: type: string description: >- `https://` URL that receives the result (`POST`, JSON). With it the request answers `202` at once. Must resolve to a public address. webhook_secret: type: string description: >- Optional, 1-256 characters: the webhook is signed with it (`X-Mihu-Signature`). webhook_metadata: type: string description: 'Optional, up to 4 KB: echoed back as `metadata`.' file_url: type: string description: >- Instead of `file`: `https://` URL the server downloads the audio from (up to 1 GB, 10 minutes, public addresses only). responses: '200': description: Transcription content: application/json: example: text: 'Adresiniz Şirinyer Mahallesi, Cumhuriyet Caddesi No: 17.' model: umay language: tr duration: 7.2 processing_time: 0.16 speakers: - speaker: S1 gender: female percent: 97 speech_seconds: 6.8 '202': description: Accepted (with `webhook_url`); the result follows on the webhook content: application/json: example: id: 6f1d2c3b-9a4e-4b7c-8d2f-1e0a5b9c7d36 object: transcription.job status: queued duration: 3600 created_at: 1793145600 source: type: upload filename: call-1042.wav size_bytes: 115200044 request: model: umay language: tr response_format: json language_model: false normalize: true diarize: true gender_detection: false speaker_lock: false voice_isolation: false hints: false max_speakers: 8 '400': description: Invalid field '401': description: Missing or wrong API key '413': description: File larger than 1 GB or audio longer than 4 hours '422': description: Invalid value or unusable audio '429': description: More than 20 jobs of this API key waiting or running '503': description: Server busy; retry later components: securitySchemes: bearerAuth: type: http scheme: bearer description: Your API key ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.