induslabs 0.0.12

Python SDK for IndusLabs Voice API - Text-to-Speech and Speech-to-Text

IndusLabs Python SDK

Official Python SDK for IndusLabs Voice API - providing Text-to-Speech (TTS) and Speech-to-Text (STT) capabilities.

Installation

pip install induslabs

Quick Start

from induslabs import Client

# Initialize client
client = Client(api_key="your_api_key_here")
# Or use environment variable: export INDUSLABS_API_KEY="your_api_key_here"
client = Client()

# Text-to-Speech
response = client.tts.speak(
    text="नमस्ते, यह एक टेस्ट है",
    voice="urvashi",
    language="hi-IN"
)
response.save("output.wav")

# Speech-to-Text
result = client.stt.transcribe("audio.wav", language="hi")
print(result.text)

Features

• ✅ Synchronous and Asynchronous APIs - Use sync methods or async for better performance
• ✅ Streaming Support - Start playing audio as soon as first bytes arrive
• ✅ Multiple Audio Formats - Support for WAV, MP3, and PCM formats
• ✅ Concurrent Requests - Built-in support for making multiple requests simultaneously
• ✅ File-like Objects - Work with audio in memory without saving to disk
• ✅ Type Hints - Full type annotations for better IDE support
• ✅ Comprehensive Error Handling - Clear error messages and exceptions

Text-to-Speech (TTS)

Basic Usage

from induslabs import Client

client = Client(api_key="your_api_key")

# Simple speech synthesis
response = client.tts.speak(
    text="Hello, this is a test",
    voice="urvashi"
)
response.save("output.wav")

Streaming Audio

Stream audio and start playing immediately:

# Enable streaming
response = client.tts.speak(
    text="This is streaming audio",
    voice="urvashi",
    stream=True
)

# Iterate over audio chunks as they arrive
for chunk in response.iter_bytes(chunk_size=8192):
    # Play or process chunk immediately
    audio_player.play(chunk)

# Or save to file
response.save("output.wav")

Working with File Objects

No need to save to disk - work with audio in memory:

response = client.tts.speak(text="Test audio", voice="urvashi")

# Get as file-like object
audio_file = response.to_file_object()

# Get raw bytes
audio_bytes = response.get_audio_data()

# Access metadata
print(f"Sample Rate: {response.sample_rate}Hz")
print(f"Channels: {response.channels}")
print(f"Bit Depth: {response.bit_depth}")

Different Audio Formats

# WAV format (default)
wav_response = client.tts.speak(text="Test", voice="urvashi", output_format="wav")

# MP3 format
mp3_response = client.tts.speak(text="Test", voice="urvashi", output_format="mp3")

# PCM format
pcm_response = client.tts.speak(text="Test", voice="urvashi", output_format="pcm")

Advanced Options

response = client.tts.speak(
    text="Advanced TTS example",
    voice="urvashi",
    language="hi-IN",
    output_format="wav",
    stream=True,
    model="orpheus-3b",  # default model
    temperature=0.7,      # optional: control randomness
    max_tokens=2000       # optional: limit generation
)

Speech-to-Text (STT)

Basic Usage

from induslabs import Client

client = Client(api_key="your_api_key")

# Transcribe audio file
result = client.stt.transcribe("audio.wav", language="hi")
print(result.text)
print(f"Detected language: {result.language_detected}")
print(f"Duration: {result.audio_duration_seconds}s")

From File-like Objects

# From bytes
with open("audio.wav", "rb") as f:
    result = client.stt.transcribe(f, language="hi")
    print(result.text)

# From BytesIO
from io import BytesIO
audio_buffer = BytesIO(audio_bytes)
result = client.stt.transcribe(audio_buffer, language="hi")

Advanced Options

result = client.stt.transcribe(
    file="audio.wav",
    language="hi",
    chunk_length_s=6,    # chunk length in seconds
    stride_s=5.9,        # stride length in seconds
    overlap_words=7      # number of overlapping words
)

# Access detailed information
print(f"Text: {result.text}")
print(f"Processing time: {result.processing_time_seconds}s")
print(f"First token time: {result.first_token_time_seconds}s")
print(f"Credits used: {result.credits_used}")
print(f"Request ID: {result.request_id}")

# Get raw response
raw_data = result.to_dict()

Async API

For better performance with concurrent requests:

import asyncio
from induslabs import Client

async def main():
    client = Client(api_key="your_api_key")
    
    # Async TTS
    response = await client.tts.speak_async(
        text="Async speech synthesis",
        voice="urvashi",
        stream=True
    )
    
    # Async iteration over chunks
    async for chunk in response.iter_bytes():
        # Process chunk
        pass
    
    # Async STT
    result = await client.stt.transcribe_async("audio.wav", language="hi")
    print(result.text)
    
    # Clean up
    await client.close()

# Run
asyncio.run(main())

Async Context Manager

async def main():
    async with Client(api_key="your_api_key") as client:
        response = await client.tts.speak_async(text="Test", voice="urvashi")
        result = await client.stt.transcribe_async("audio.wav")
        # Auto cleanup on exit

asyncio.run(main())

Concurrent Requests

Sync Concurrent Requests

from concurrent.futures import ThreadPoolExecutor
from induslabs import Client

client = Client(api_key="your_api_key")

texts = ["Text 1", "Text 2", "Text 3", "Text 4"]

def generate_speech(text):
    return client.tts.speak(text=text, voice="urvashi")

# Generate multiple audio files concurrently
with ThreadPoolExecutor(max_workers=4) as executor:
    responses = list(executor.map(generate_speech, texts))

for i, response in enumerate(responses):
    response.save(f"output_{i}.wav")

Async Concurrent Requests

import asyncio
from induslabs import Client

async def main():
    client = Client(api_key="your_api_key")
    
    texts = ["Text 1", "Text 2", "Text 3", "Text 4"]
    
    # Create tasks
    tasks = [
        client.tts.speak_async(text=text, voice="urvashi")
        for text in texts
    ]
    
    # Run concurrently
    responses = await asyncio.gather(*tasks)
    
    # Save all
    for i, response in enumerate(responses):
        response.save(f"output_{i}.wav")
    
    await client.close()

asyncio.run(main())

Error Handling

from induslabs import Client
import requests

client = Client(api_key="your_api_key")

try:
    response = client.tts.speak(text="Test", voice="urvashi")
    response.save("output.wav")
except requests.exceptions.HTTPError as e:
    print(f"HTTP error occurred: {e}")
except ValueError as e:
    print(f"Invalid parameter: {e}")
except Exception as e:
    print(f"An error occurred: {e}")

Available Voices

Currently supported voices:

• urvashi (Hindi female)
• More voices coming soon!

Supported Languages

The API supports multiple languages. Specify using language codes:

• hi or hi-IN - Hindi
• en or en-US - English
• And more...

Audio Format Details

WAV (default)

• Sample Rate: 24000 Hz
• Channels: 1 (mono)
• Bit Depth: 16-bit
• Best for: Quality and compatibility

MP3

• Compressed format
• Best for: Smaller file sizes

PCM

• Raw audio data
• Best for: Direct audio processing

Environment Variables

# Set API key
export INDUSLABS_API_KEY="your_api_key_here"

# Then use without passing api_key
client = Client()

Requirements

• Python >= 3.7
• requests >= 2.25.0
• aiohttp >= 3.8.0

Support

• Documentation: https://docs.induslabs.io
• Issues: https://github.com/INDUS-AI-DEV/induslabs-python/issues
• Email: [e-mail]

License

MIT License - see LICENSE file for details.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.