Speech-to-Text API with Voice Profiling: Emotion, Accent, Age, and Pitch Detection

import requests
import base64
import json

# pip install requests
API_KEY = "YOUR_API_KEY"  # From https://platform.inworld.ai

STT_ENDPOINT = "https://api.inworld.ai/stt/v1/transcribe"

# Read and base64-encode the audio file
with open("call_recording.wav", "rb") as audio_file:
    audio_b64 = base64.b64encode(audio_file.read()).decode("utf-8")

# Transcribe with Inworld STT
# Voice profiling features (emotion, accent, age, pitch, style) are available
# with the Inworld STT model. Check https://docs.inworld.ai/stt/overview
# for the latest supported features and response fields.
response = requests.post(
    STT_ENDPOINT,
    headers={
        "Authorization": f"Basic {API_KEY}",
        "Content-Type": "application/json"
    },
    json={
        "transcribeConfig": {
            "modelId": "inworld/inworld-stt-1",
            "audioEncoding": "AUTO_DETECT",
            "language": "en-US"
        },
        "audioData": {
            "content": audio_b64
        }
    },
    timeout=60
)
response.raise_for_status()

result = response.json()

# Access the transcript
transcription = result["transcription"]
print("Transcript:", transcription["transcript"])
print("Is final:", transcription["isFinal"])

# Access word-level timestamps
for word in transcription.get("wordTimestamps", []):
    print(f"  '{word['word']}' at {word['startTime']}-{word['endTime']}")

# Voice profiling fields (emotion, accent, age, pitch, style) may be available
# with the inworld/inworld-stt-1 model. See STT docs for current schema.
print("Usage:", result.get("usage", {}))

import { readFile } from 'node:fs/promises';

const STT_ENDPOINT = 'https://api.inworld.ai/stt/v1/transcribe';
const API_KEY = 'YOUR_API_KEY'; // From https://platform.inworld.ai

async function transcribeWithProfiling(audioPath) {
  const audioBuffer = await readFile(audioPath);
  const audioB64 = audioBuffer.toString('base64');

  // Voice profiling features (emotion, accent, age, pitch, style) are available
  // with the Inworld STT model. Check https://docs.inworld.ai/stt/overview
  // for the latest supported features and response fields.
  const response = await fetch(STT_ENDPOINT, {
    method: 'POST',
    headers: {
      'Authorization': `Basic ${API_KEY}`,
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      transcribeConfig: {
        modelId: 'inworld/inworld-stt-1',
        audioEncoding: 'AUTO_DETECT',
        language: 'en-US',
      },
      audioData: {
        content: audioB64,
      },
    }),
  });

  if (!response.ok) {
    throw new Error(`STT API error ${response.status}: ${await response.text()}`);
  }

  const result = await response.json();

  console.log('Transcript:', result.transcription.transcript);
  console.log('Is final:', result.transcription.isFinal);

  // Access word-level timestamps
  for (const word of result.transcription.wordTimestamps || []) {
    console.log(`  '${word.word}' at ${word.startTime}-${word.endTime}`);
  }

  // Voice profiling fields (emotion, accent, age, pitch, style) may be available
  // with the inworld/inworld-stt-1 model. See STT docs for current schema.
  console.log('Usage:', result.usage);

  return result;
}

transcribeWithProfiling('call_recording.wav');