Connecting…
Pakistan ki apni Urdu Speech-to-Text

بولیں،
تحریر پائیں

On premise Urdu speech recognition, banaya gaya Pakistani zameen par. Mic se record karein ya audio file upload karein — saaf Urdu transcript wapas milega. Sab kuch aapke apne infrastructure par.

اپنی پہچان، اپنی آواز
آواز داخل کریں
0:00
Mic button dabayein aur bolna shuru karein
Audio file yahan drop karein ya click karein
WAV · MP3 · M4A · OGG · WEBM · FLAC — max 25 MB
🎧
recording.webm
Transcribe ke liye tayyar
Transcript
Language · Duration · Words ·
Transcription history
No transcriptions yet

Overview

Callrolin STT API provides on-premise Urdu speech-to-text. Send an audio file and receive a clean transcript in a JSON response. Also supports Punjabi, Sindhi, Pashto and English, plus translation to English.

Base URL
http://34.93.31.201:8000

Requests use multipart/form-data (audio upload) and return an application/json transcript. Authentication is required for all /v1/ endpoints.

Authentication

Authenticate your requests by including your API key in the X-API-Key header.

Enter your API key above and all code examples will update automatically. Contact your administrator to get an API key.
HeaderValueRequired
X-API-KeyYour secret API keyRequired
Content-Typemultipart/form-dataRequired

Alternatively, pass the key as a Bearer token:

Authorization: Bearer smat-your-api-key

POST /v1/stt/transcribe

Convert an audio file to text. Returns a JSON transcript.

POST http://34.93.31.201:8000/v1/stt/transcribe

Form fields

FieldTypeRequiredDefaultDescription
filebinaryRequired·Audio file. WAV, MP3, M4A, OGG, WEBM, FLAC. Max 25 MB.
modelstringRequired·Model ID. Must be samaat-v1
languagestringOptionalurISO code: ur, pa, sd, ps, en, or auto.
taskstringOptionaltranscribetranscribe or translate (to English).

Response

Returns application/json:

{ "text": "پاکستان ایک خوبصورت ملک ہے", "language": "ur", "duration": 4.12, "model": "samaat-v1", "request_id": "req_a3f8b2c1d4e5" }

Code Examples

Models

Currently one model is available. Additional models will be released in future versions.

Model IDLanguagesSample RateDescription
samaat-v1 Urdu, Punjabi, Sindhi, Pashto, English 16,000 Hz On-premise Urdu-first ASR. Handles noisy phone audio and code-switching. Optional English translation.

List Models

GET http://34.93.31.201:8000/v1/models
curl http://34.93.31.201:8000/v1/models

Error Codes

The API uses standard HTTP status codes. Error responses include a JSON body with an error object.

{ "error": { "code": "file_too_large", "message": "Audio exceeds 25 MB limit.", "type": "invalid_request_error" } }
HTTP StatusError CodeCause
401missing_api_keyNo API key provided in request headers
401invalid_api_keyAPI key is incorrect or has been deactivated
400invalid_modelModel ID is not recognized
400no_audio_fileNo file field in the request
413file_too_largeAudio exceeds the 25 MB limit
415unsupported_mediaAudio format could not be decoded
500inference_errorInternal model error during transcription