Speech to Text › Diarization
Diarization
Identify and label the speakers in multi-speaker Arabic audio, then get each speaker segment aligned with transcribed text and timestamps — who said what in meetings, interviews, podcasts and conversations.
1
Endpoint
POST /api/v1/audio/diarization/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. How it works: you upload a multi-speaker file, Munsit identifies speaker turns and assigns speaker labels, and you receive transcription, diarization segments, and merged speaker-labeled text with timing.
| 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). |
Transcribe-only options.
hotwords, language and return_confidence apply to POST /audio/transcribe only.Only need to know who said what? Add
diarize=true to POST /audio/transcribe and the same transcript comes back with a speaker number on every word and the speech grouped into utterances. This endpoint adds the speaker segments (diarization) and the speaker-labeled text merged with them (merged), and its results can feed Diarization + sentiment.Pricing. 110 credits per minute of audio: the transcription base rate plus speaker labels, the same as
diarize=true on POST /audio/transcribe. See Cost calculation.3
Example request
Works like Transcribe, with speaker labels added to the response.
curl -X POST "https://api.munsit.com/api/v1/audio/diarization/transcribe" \ -H "x-api-key: $MUNSIT_API_KEY" \ -F "file=@interview.mp3" \ -F "model=munsit"
import requests, osr = requests.post( "https://api.munsit.com/api/v1/audio/diarization/transcribe", headers={"x-api-key": os.environ["MUNSIT_API_KEY"]}, files={"file": open("interview.mp3", "rb")}, data={"model": "munsit"},)for seg in r.json()["data"]["merged"]: print(seg["speaker"], seg["text"])
import fs from "node:fs";const form = new FormData();form.append("file", new Blob([fs.readFileSync("interview.mp3")]), "interview.mp3");form.append("model", "munsit");const res = await fetch( "https://api.munsit.com/api/v1/audio/diarization/transcribe", { method: "POST", headers: { "x-api-key": process.env.MUNSIT_API_KEY }, body: form });const { data } = await res.json();for (const seg of data.merged) console.log(seg.speaker, seg.text);
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("interview.mp3") fw, _ := w.CreateFormFile("file", "interview.mp3") io.Copy(fw, f) w.WriteField("model", "munsit") w.Close() req, _ := http.NewRequest("POST", "https://api.munsit.com/api/v1/audio/diarization/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) for _, seg := range data["merged"].([]any) { s := seg.(map[string]any) fmt.Println(s["speaker"], s["text"]) }}
Response — 200
{
"statusCode": 200,
"data": {
"transcription": {
"transcription": "السلام عليكم ورحمة الله وبركاته. كيف حالك اليوم؟",
"timestamps": [
{ "word": "السلام", "start": 0.0, "end": 1.2 }
]
},
"diarization": {
"segments": [
{ "start": 0.0, "end": 8.5, "speaker": "SPEAKER_00" }
]
},
"merged": [
{
"start": 0.0,
"end": 8.5,
"speaker": "SPEAKER_00",
"text": "السلام عليكم ورحمة الله وبركاته"
}
],
"duration": 53.661375
},
"message": "Success"
}
4
Response fields
Three views of the same audio arrive under data: the raw transcript, the speaker segments, and the two merged together.
| Field | Type | Description |
|---|---|---|
transcription | object | Transcript and word-level timestamps. |
diarization | object | Speaker segments with start/end labels. |
merged | array of objects (start, end, speaker, text) | Combined diarization + transcript. |
duration | number | Total duration in seconds. |
transcriptionId | string (UUID) | Transcription identifier. Use it as the path parameter for diarization sentiment analysis. |
originalTranscript | string | Raw verbatim transcript. |
attributes | object | Internal metadata blob persisted with the transcription. Prefer the named fields — treat this as unstable. |
stats | object (fileName, fileSize, mimeType, creditsConsumed) | Upload metadata and the credits billed for this request. |
Most apps only need
merged. Each entry is one speaker turn — speaker, its text, and start/end timing — ready to render as a conversation.
5
Go further
What to run on a diarized conversation next.
Diarization + sentiment →Analyze per-speaker sentiment from a diarization record.POST /diarization/{diarizationId}/sentiment-analysis
Transcribe →Plain transcription when you don't need speaker labels.POST /audio/transcribe
Minutes of meetings →Structured meeting transcripts for documentation.POST /minutes-of-meeting/transcribe
Keyword extraction →Pull the key topics out of the conversation.POST /keyword-extraction
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.