Skip to main content
CallMissed
DocumentationAPI ReferencesCookbook

Speech to Text

BATCH STT

Batch speech-to-text transcription with speaker diarization for call analytics.

Overview

Batch STT with speaker diarization is available through the Call Analytics API. Upload audio files and receive:

  • Diarized transcript (SPEAKER_1, SPEAKER_2, etc.)
  • Per-speaker timing breakdown
  • LLM-powered analysis across 9 dimensions
  • Structured summaries

Supported formats: WAV, MP3, MP4, M4A, OGG, FLAC, WebM, AAC, AMR (max 100 MB)

Analyze a Recording

bash
POST /api/v1/analytics/calls/analyze
Authorization: Bearer <access_token>
Content-Type: multipart/form-data
import requests

response = requests.post(
    "https://api.callmissed.com/api/v1/analytics/calls/analyze",
    headers={"Authorization": "Bearer YOUR_ACCESS_TOKEN"},
    files={"file": open("call_recording.wav", "rb")},
    data={"language": "hi-IN"},
)

result = response.json()
print(result["transcript"])
print("Speaker timing:", result["speaker_timing"])
print("Analysis:", result["analysis"])
FieldTypeRequiredDescription
filefileYesAudio file (max 100 MB)
languagestringNoBCP-47 language code (default: unknown for auto-detect)

Ask Questions

After analyzing a call, ask follow-up questions about the transcript:

response = requests.post(
    "https://api.callmissed.com/api/v1/analytics/calls/question",
    headers={
        "Authorization": "Bearer YOUR_ACCESS_TOKEN",
        "Content-Type": "application/json",
    },
    json={
        "transcript": result["transcript"],
        "question": "Did the customer agree to the terms?"
    },
)

print(response.json()["answer"])

Response Format

json
{
  "transcript": "SPEAKER_1: Namaste, main aapki kaise madad kar sakta hoon?\nSPEAKER_2: Mujhe apne order ke baare mein jaanna hai.",
  "speaker_timing": {
    "SPEAKER_1": 45.2,
    "SPEAKER_2": 38.7
  },
  "analysis": "Customer Sentiment: Neutral\nResolution: Resolved\nAgent Performance: Good\n...",
  "summary": "Order inquiry, resolved successfully",
  "generated_at": "2026-04-12T10:00:00Z"
}

The analysis covers 9 dimensions: sentiment, intent, resolution, agent performance, escalation risk, key topics, action items, compliance, and satisfaction score.

See the Call Analytics Cookbook for a full walkthrough with real examples.

Previous

Real-time

Next

Overview

Was this page helpful?