Speech to TextFeatures

Speaker diarization

View as MarkdownOpen in Claude
Pre-Recorded Real-Time

Enabling speaker diarization

Pre-Recorded API

Pass diarize=true when calling the Pulse STT POST endpoint. The parameter can be combined with other enrichment options (timestamps, emotions, etc.) without changing your audio payload.

$curl --request POST \
>  --url "https://waves-api.smallest.ai/api/v1/pulse/get_text?model=pulse&language=en&diarize=true" \
>  --header "Authorization: Bearer $SMALLEST_API_KEY" \
>  --header "Content-Type: audio/wav" \
>  --data-binary "@/path/to/audio.wav"

Real-Time WebSocket API

Add diarize=true to your WebSocket connection query parameters when connecting to the Pulse STT WebSocket API.

1const url = new URL("wss://waves-api.smallest.ai/api/v1/pulse/get_text");
2url.searchParams.append("language", "en");
3url.searchParams.append("encoding", "linear16");
4url.searchParams.append("sample_rate", "16000");
5url.searchParams.append("diarize", "true");
6
7const ws = new WebSocket(url.toString(), {
8  headers: {
9    Authorization: `Bearer ${API_KEY}`,
10  },
11});

Output format & field of interest

When enabled, every entry in words includes a speaker field (integer ID: 0, 1, …) and speaker_confidence field (0.0 to 1.0) for real-time API, or string labels (speaker_0, speaker_1, …) for pre-recorded API. The utterances array also carries speaker labels so you can reconstruct conversations, build turn-taking analytics, or display multi-speaker captions.

Pre-Recorded API

Sample request

$curl --request POST \
>  --url "https://waves-api.smallest.ai/api/v1/pulse/get_text?model=pulse&language=en&diarize=true" \
>  --header "Authorization: Bearer $SMALLEST_API_KEY" \
>  --header "Content-Type: audio/wav" \
>  --data-binary "@/path/to/two-speaker.wav"

Sample response

Pre-Recorded API Response

1{
2  "transcription": "Agent: Hello world. Customer: Hi there.",
3  "words": [
4    { "start": 0.0, "end": 0.4, "speaker": "speaker_0", "word": "Hello" },
5    { "start": 0.4, "end": 0.8, "speaker": "speaker_0", "word": "world." },
6    { "start": 1.0, "end": 1.2, "speaker": "speaker_1", "word": "Hi" },
7    { "start": 1.2, "end": 1.6, "speaker": "speaker_1", "word": "there." }
8  ],
9  "utterances": [
10    { "text": "Hello world.", "start": 0.0, "end": 0.8, "speaker": "speaker_0" },
11    { "text": "Hi there.", "start": 1.0, "end": 1.6, "speaker": "speaker_1" }
12  ]
13}

Real-Time WebSocket API Response

1{
2  "session_id": "sess_12345abcde",
3  "transcript": "Hello world. Hi there.",
4  "is_final": true,
5  "is_last": false,
6  "language": "en",
7  "words": [
8    {
9      "word": "Hello",
10      "start": 0.0,
11      "end": 0.4,
12      "confidence": 0.98,
13      "speaker": 0,
14      "speaker_confidence": 0.95
15    },
16    {
17      "word": "world.",
18      "start": 0.4,
19      "end": 0.8,
20      "confidence": 0.97,
21      "speaker": 0,
22      "speaker_confidence": 0.92
23    },
24    {
25      "word": "Hi",
26      "start": 1.0,
27      "end": 1.2,
28      "confidence": 0.99,
29      "speaker": 1,
30      "speaker_confidence": 0.88
31    },
32    {
33      "word": "there.",
34      "start": 1.2,
35      "end": 1.6,
36      "confidence": 0.96,
37      "speaker": 1,
38      "speaker_confidence": 0.91
39    }
40  ],
41  "utterances": [
42    {
43      "text": "Hello world.",
44      "start": 0.0,
45      "end": 0.8,
46      "speaker": 0
47    },
48    {
49      "text": "Hi there.",
50      "start": 1.0,
51      "end": 1.6,
52      "speaker": 1
53    }
54  ]
55}

Response Fields

FieldTypeWhen IncludedDescription
speakerinteger (realtime) / string (pre-recorded)diarize=trueSpeaker label. Real-time API uses integer IDs (0, 1, …), pre-recorded API uses string labels (speaker_0, speaker_1, …)
speaker_confidencenumberdiarize=true (realtime only)Confidence score for the speaker assignment (0.0 to 1.0)