Speech to Text › Transcribe
Transcribe
Upload pre-recorded Arabic audio as multipart/form-data and receive a high-quality transcript with total duration and word-level timestamps. Optimized for asynchronous processing of interviews, meetings, media clips and customer calls.
1
Endpoint
POST /api/v1/audio/transcribe
Authenticate with your API key in the x-api-key header. See Authentication.
| Header | Value |
|---|---|
x-api-key | YOUR_MUNSIT_API_KEY |
2
Request
Send the body as multipart/form-data.
| Field | Type | Required | Description |
|---|---|---|---|
file | file | Yes | Audio file in a supported format — see what you can send. |
model | string | No | ASR model to use: munsit (default) or munsit-en-ar (mixed Arabic-English with code-switching). |
hotwords | string | No | Comma-separated custom vocabulary (multi-word phrases allowed). Biases recognition toward names, brands and domain terms. Not supported with munsit-en-ar. |
return_confidence | boolean | No | When true, each timestamps entry includes a confidence score (0–1). The response is otherwise unchanged. |
return_timestamps | boolean | No | Defaults to true on munsit; set false to return an empty timestamps array. On munsit-en-ar timestamps are off unless you set this to true. |
return_turns | boolean | No | When true, adds the turns array and tags each turn with smart-turn is_complete and turn_probability. These two fields appear only with this flag. |
return_gender | boolean | No | When true, adds the turns array with a gender object (label, score) per turn, plus a whole-file rollup in analysis. |
return_sentiment | boolean | No | When true, adds the turns array with a sentiment object (label, score) per turn, plus a whole-file rollup in analysis. |
3
Custom vocabulary
Pass rare terms the recognizer is unlikely to know — customer names, product codes, brand words. In our benchmarks, biasing recovered rare terms from 0% to 77% recall with overall accuracy unchanged or slightly better. Short lists of 5–30 genuinely rare terms work best; very long lists dilute the effect.
curl -X POST "https://api.munsit.com/api/v1/audio/transcribe" \
-H "x-api-key: $MUNSIT_API_KEY" \
-F "file=@call.wav" \
-F "hotwords=عبد القادر,أديب" \
-F "return_confidence=true"
Response — timestamps entry
{ "word": "الأشياء", "start": 0.24, "end": 0.31, "confidence": 0.994 }
hotwords is ignored when model=munsit-en-ar. The mixed Arabic-English model does not support custom vocabulary.Transcribe only.
hotwords and return_confidence apply to POST /audio/transcribe — not to diarization or minutes of meetings. For live audio, streaming takes hotwords as a query parameter and always returns confidence.4
Per-turn analysis
Set return_turns, return_gender and/or return_sentiment to break the transcript into turns and annotate each one. Any of the three adds the turns array — the gender and sentiment flags imply it — and each flag contributes only its own fields. analysis holds the whole-file rollup and appears only with return_gender or return_sentiment; return_turns on its own does not produce it. Short single-speaker recordings typically come back as one turn.
curl -X POST "https://api.munsit.com/api/v1/audio/transcribe" \
-H "x-api-key: $MUNSIT_API_KEY" \
-F "file=@call.wav" \
-F "return_turns=true" \
-F "return_gender=true" \
-F "return_sentiment=true"
Response — 200
{
"statusCode": 200,
"data": {
"transcription": "أهلا كيف حالك... بخير شكرا",
"duration": 6.2,
"turns": [
{
"turn_id": 0,
"start": 0.0,
"end": 2.8,
"text": "أهلا كيف حالك",
"is_complete": true,
"turn_probability": 0.94,
"gender": { "label": "male", "score": 0.88 },
"sentiment": { "label": "neutral", "score": 0.81 }
},
{
"turn_id": 1,
"start": 3.1,
"end": 6.2,
"text": "بخير شكرا",
"is_complete": true,
"turn_probability": 0.97,
"gender": { "label": "female", "score": 0.91 },
"sentiment": { "label": "positive", "score": 0.86 }
}
],
"analysis": {
"turns": 2,
"gender": { "dominant": "female", "by_duration_s": { "male": 2.8, "female": 3.1 } },
"sentiment": { "dominant": "positive", "counts": { "neutral": 1, "positive": 1 } }
}
},
"message": "Success"
}
This sentiment is a quick per-utterance signal returned alongside the transcript. For deeper, LLM-based analysis of an existing transcription — emotions, trends, critical moments — use Sentiment analysis instead. For live audio, streaming emits
Sentiment and Gender events per turn.GET /history/speech-to-text/:id also returns turns and analysis for transcriptions created with one of these flags set.5
Example request
How it works: upload audio, Munsit analyzes the recording and converts the Arabic speech into text, and you get back a transcript with duration and word-level timestamps.
curl -X POST "https://api.munsit.com/api/v1/audio/transcribe" \ -H "x-api-key: $MUNSIT_API_KEY" \ -F "file=@meeting.mp3" \ -F "model=munsit"
import requests, osr = requests.post( "https://api.munsit.com/api/v1/audio/transcribe", headers={"x-api-key": os.environ["MUNSIT_API_KEY"]}, files={"file": open("meeting.mp3", "rb")}, data={"model": "munsit"},)out = r.json()["data"]print(out["transcription"], out["duration"])
import fs from "node:fs";const form = new FormData();form.append("file", new Blob([fs.readFileSync("meeting.mp3")]), "meeting.mp3");form.append("model", "munsit");const res = await fetch( "https://api.munsit.com/api/v1/audio/transcribe", { method: "POST", headers: { "x-api-key": process.env.MUNSIT_API_KEY }, body: form });const { data } = await res.json();console.log(data.transcription, data.duration);
package mainimport ( "bytes" "encoding/json" "fmt" "io" "mime/multipart" "net/http" "os")func main() { var buf bytes.Buffer w := multipart.NewWriter(&buf) f, _ := os.Open("meeting.mp3") fw, _ := w.CreateFormFile("file", "meeting.mp3") io.Copy(fw, f) w.WriteField("model", "munsit") w.Close() req, _ := http.NewRequest("POST", "https://api.munsit.com/api/v1/audio/transcribe", &buf) req.Header.Set("x-api-key", os.Getenv("MUNSIT_API_KEY")) req.Header.Set("Content-Type", w.FormDataContentType()) resp, _ := http.DefaultClient.Do(req) defer resp.Body.Close() var out map[string]any json.NewDecoder(resp.Body).Decode(&out) data := out["data"].(map[string]any) fmt.Println(data["transcription"], data["duration"])}
Response — 200
{
"statusCode": 200,
"data": {
"transcriptionId": "805059bf-7c3f-4a1e-9d2b-1f0c6ae83b47",
"transcription": "لك كلما عمقت الآخرين أصبحت قزما...",
"duration": 53.661375,
"timestamps": [
{ "word": "الأشياء", "start": 0.24, "end": 0.31 }
],
"summary": "",
"audioUrl": "<stored-audio-url>",
"stats": {
"fileName": "meeting.mp3",
"fileSize": "1.42 MB",
"mimeType": "audio/mpeg",
"creditsConsumed": 7
}
},
"message": "Success"
}
Files must be under 60 minutes. For longer recordings, split the audio into shorter segments for best performance — or use Streaming, which has no duration limit.
6
Response fields
The payload arrives under data, alongside statusCode and message.
| Field | Type | Description |
|---|---|---|
transcriptionId | string (UUID) | Transcription identifier. Pass it as the path parameter to sentiment analysis and keyword extraction. |
transcription | string | Full transcript text. |
duration | number | Audio duration in seconds. |
timestamps | array of objects (word, start, end, plus confidence when return_confidence=true) | Word-level timestamps. |
turns | array of objects (turn_id, start, end, text; plus is_complete / turn_probability with return_turns, gender with return_gender, sentiment with return_sentiment) | Present when any of the three flags is true — return_gender and return_sentiment imply it. Short single-speaker clips typically return one turn. |
analysis | object (turns count; plus gender and sentiment when those flags are set) | Whole-file rollup. Present only with return_gender or return_sentiment — return_turns alone does not produce it. |
attributes | object | Internal metadata blob persisted with the transcription; mirrors fields above under different names (and repeats the full timestamps array). Prefer the named fields — treat this as unstable. |
summary | string | Always present. Empty string on plain transcription; carries the generated summary on minutes of meetings. |
audioUrl | string | Stored copy of the uploaded audio. See Audio retention. |
stats | object (fileName, fileSize, mimeType, creditsConsumed) | Upload metadata and the credits billed for this request. |
7
Audio retention
Audio submitted to the hosted API is stored server-side. Every successful response carries an audioUrl pointing at that stored copy, and the transcript text is persisted alongside it so it can be retrieved later through the history endpoints.
| What | Detail |
|---|---|
| What is stored | The uploaded audio file and the resulting transcript, keyed to your account. |
| Access | The response returns an audioUrl for the stored object. Do not treat it as a secret or hard-code it — the exact form is not part of the API contract and may change. |
| Removal | Delete a transcription and its stored audio through the history endpoints, or contact support for bulk removal. |
| Avoiding retention | If your deployment cannot retain audio at all, use self-hosting, where storage is under your control. |
8
Go further
Do more with your transcript.
Streaming →Live transcripts over a WebSocket instead of file upload.WSS /api/v1/listen
Diarization →Speaker-labeled segments merged with the transcript.POST /audio/diarization/transcribe
Minutes of meetings →Structured transcripts for meeting records.POST /minutes-of-meeting/transcribe
Sentiment analysis →Analyze the tone of what was transcribed.POST /sentiment-analysis
Working with an AI assistant? Every page is available as Markdown: add .md to the URL, or send an Accept: text/markdown header. For the whole documentation in one request, point it at llms-full.txt; the page index is llms.txt. Or use Copy Page, top right.