Skip to main content
Text to speech

Quickstart

Convert text to natural-sounding speech with our text to speech API.

Transform written text into high-quality, natural-sounding speech using our advanced text to speech API.

Text to speech is currently in free beta. Features and pricing may change as we continue to improve the service. We welcome your feedback to help shape the final product.

Use Cases​

Our text to speech API is perfect for:

  • Voice AI: Voice assistants, chatbots, and IVR systems
  • Accessibility applications: Screen readers and assistive technologies
  • Content creation: Podcasts, audiobooks, and voice-overs
  • E-learning platforms: Course narration and educational content
  • Gaming: Character voices and dynamic dialogue
  • Media production: News broadcasts and automated announcements

Getting Started​

1. Get your API key​

Create an API key in the portal to access the Text-to-Speech API. Store your key securely as a managed secret.

2. Install dependencies​

pip install requests numpy

3. Make your first request​

Replace YOUR_API_KEY with your actual API key from the portal:

import requests
import wave
import numpy as np

url = "https://preview.tts.speechmatics.com/generate"
headers = {
    "Authorization": "Bearer YOUR_API_KEY"
}
data = {
    "text": "Hello world! This is my first text-to-speech conversion."
}

response = requests.post(url, headers=headers, json=data)

if response.status_code == 200:
    # Get raw PCM data (32-bit float, little-endian)
    raw_audio = response.content
    
    # Convert raw bytes to numpy array of 32-bit floats (little-endian)
    float_samples = np.frombuffer(raw_audio, dtype='<f4')  # '<f4' = little-endian 32-bit float
    
    # Convert float samples (-1.0 to 1.0) to 16-bit integers for WAV compatibility
    # Scale to 16-bit range and convert to integers
    int16_samples = (float_samples * 32767).astype(np.int16)
    
    # WAV file parameters
    sample_rate = 16000  # 16 kHz
    channels = 1         # Mono
    sample_width = 2     # 16-bit = 2 bytes (for compatibility)
    
    # Create WAV file
    with wave.open("output.wav", "wb") as wav_file:
        wav_file.setnchannels(channels)
        wav_file.setsampwidth(sample_width)
        wav_file.setframerate(sample_rate)
        wav_file.writeframes(int16_samples.tobytes())
    
    duration = len(float_samples) / sample_rate
    print(f"Speech synthesis completed! Saved {len(raw_audio)} bytes as output.wav")
    print(f"Converted {len(float_samples)} float32 samples to 16-bit WAV")
    print(f"Duration: {duration:.2f} seconds")
else:
    print(f"Error: {response.status_code}")
    print(f"Response: {response.text}")

Output Format​

The Text-to-Speech API returns raw PCM audio data with the following specifications:

  • Format: Raw PCM (Pulse Code Modulation)
  • Sample Rate: 16 kHz
  • Bit Depth: 32-bit
  • Channels: Mono (single channel)
  • Byte Order: Little-endian

Best Practices​

  • Use proper punctuation: Periods, commas, and question marks improve natural speech flow
  • Spell out abbreviations: Write "Doctor" instead of "Dr." for clearer pronunciation
  • Batch requests: Combine multiple short texts into longer requests when possible
  • Cache audio: Store generated audio files to avoid repeated API calls for identical text

Next Steps​

Ready to build something amazing with text to speech? We'd love to hear about your project and help you succeed.

Get in touch with us:

  • Share your feedback and feature requests
  • Ask questions about implementation
  • Discuss enterprise pricing and custom voices
  • Report any issues or bugs you encounter

Contact our team or join our developer community to connect with other builders using text to-Speech.