whissle 0.0.1

Python SDK for Whissle API.

Whissle API Python Package

whissle is a Python package that provides easy access to the Whissle API for Speech-to-Text (STT), Machine Translation, and Text Summarization functionalities, with both synchronous and asynchronous client support.

Features

• 🎙️ Speech-to-Text (STT) with multiple model support and customization options
• 🌍 Machine Translation across various languages
• 📝 Text Summarization using LLM models
• ⚡ Both synchronous and asynchronous clients
• 🔄 Automatic response parsing to Pydantic models

Installation

Install Whissle using pip:

pip install whissle

For local development:

pip install -e .

Authentication

Provide your authentication token in one of two ways:

1. Direct initialization:

from whissle import WhissleClient

client = WhissleClient(auth_token="your_auth_token_here")

2. Environment variable:

export WHISSLE_AUTH_TOKEN=your_auth_token_here

Client Options

Synchronous Client

Use the sync client for straightforward, sequential operations:

from whissle import WhissleClient

client = WhissleClient().sync_client

# List ASR models
models = client.list_asr_models()

# Speech to text
response = client.speech_to_text(
    audio_file_path="audio.wav",
    model_name="en-US-0.6b",
    timestamps=True,
    boosted_lm_words=["specific", "terms"],
    boosted_lm_score=80
)

# Translation
translation = client.machine_translation(
    text="Hello, world!",
    source_language="en",
    target_language="es"
)

# Summarization
summary = client.llm_text_summarizer(
    content="Long text here...",
    model_name="openai",
    instruction="Provide a brief summary"
)

Asynchronous Client

Use the async client for concurrent operations and non-blocking I/O:

import asyncio
from whissle import WhissleClient

async def main():
    client = WhissleClient().async_client

    # List ASR models
    models = await client.list_asr_models()

    # Speech to text
    response = await client.speech_to_text(
        audio_file_path="audio.wav",
        model_name="en-US-0.6b",
        timestamps=True
    )

    # Translation
    translation = await client.machine_translation(
        text="Hello, world!",
        source_language="en",
        target_language="es"
    )

    # Summarization
    summary = await client.llm_text_summarizer(
        content="Long text here...",
        model_name="openai",
        instruction="Provide a brief summary"
    )

asyncio.run(main())

API Methods

List ASR Models

models = client.list_asr_models()  # or await client.list_asr_models()

Speech-to-Text

response = client.speech_to_text(
    audio_file_path="path/to/audio.wav",
    model_name="en-US-0.6b",
    timestamps=False,  # Optional: Include word timestamps
    boosted_lm_words=["specific", "terms"],  # Optional: Boost specific words
    boosted_lm_score=80  # Optional: Boosting score
)

Machine Translation

translation = client.machine_translation(
    text="Hello, world!",
    source_language="en",
    target_language="es"
)

Text Summarization

summary = client.llm_text_summarizer(
    content="Your long text here...",
    model_name="openai",
    instruction="Provide a brief summary"
)

Configuration

• WHISSLE_AUTH_TOKEN: Authentication token
• WHISSLE_SERVER_URL: Optional custom server URL (defaults to https://api.whissle.ai/v1)

Error Handling

The library provides consistent error handling for both sync and async clients:

from whissle import HttpError

try:
    # Whissle API calls
except HttpError as e:
    print(f"API Error: {e.status_code} - {e.message}")

Dependencies

• Python 3.8+
• httpx: For HTTP requests (both sync and async)
• pydantic: For response modeling

Type Hints

All methods include proper type hints for better IDE support:

• ASRModel: ASR model information
• STTResponse: Speech-to-text response
• MTResposne: Machine translation response
• LLMSummarizerResponse: Text summarization response

Examples

For more detailed examples, check out the examples/ directory:

• examples/sync/: Synchronous client examples
• examples/async/: Asynchronous client examples

Contributing

Contributions are welcome! Please submit pull requests or open issues on our GitHub repository.

Contact

For support or inquiries, contact: [e-mail]