pinch-sdk 0.1.1

Pinch Python SDK for real-time speech translation

Pinch Python SDK

Requirements

• Python 3.10+
• A Pinch API key in PINCH_API_KEY
  • Docs: https://www.startpinch.com/docs
  • Create/manage API keys (Portal): https://portal.startpinch.com/dashboard/developers

Authentication (PINCH_API_KEY)

The SDK reads the API key from:

1. PINCH_API_KEY environment variable (recommended)
2. A local .env file in your current working directory (optional)

Example .env file (optional):

PINCH_API_KEY=your_key_here

Install (pip)

python3 -m pip install pinch-sdk

Optional (only needed for WAV resampling support for non-16k/48k inputs):

python3 -m pip install "pinch-sdk[audio]"

Install (uv)

In a uv-managed project:

uv add pinch-sdk

Usage

Create a small script (for example translate.py) in your own project:

import asyncio

from pinch import PinchClient

async def main() -> None:
    client = PinchClient()  # reads PINCH_API_KEY

    await client.translate_file(
        input_wav_path="input.wav",
        output_wav_path="output.wav",
        transcript_path="transcript.txt",
        # defaults:
        # source_language="en-US"
        # target_language="es-ES"
        # audio_output_enabled=True
    )

if __name__ == "__main__":
    asyncio.run(main())

Run (pip environment):

export PINCH_API_KEY="..."
python3 translate.py

Run (uv environment):

export PINCH_API_KEY="..."
uv run python translate.py

Outputs:

• output.wav (if audio_output_enabled=True)
• transcript.txt

Input audio notes

• Input must be 16-bit PCM WAV
• Streaming API supports sample rates: 16000 Hz (recommended) and 48000 Hz
• WAV helpers support 16000 Hz and 48000 Hz out of the box
• Other sample rates require resampling deps: python3 -m pip install "pinch-sdk[audio]"

Repo example (for this git checkout)

This repository includes an example script and a default input WAV.

Run with pip:

python3 -m pip install -e .
export PINCH_API_KEY="..."
python3 examples/translate.py

Run with uv:

uv sync
export PINCH_API_KEY="..."
uv run python examples/translate.py

Outputs (generated):

• examples/output.wav
• examples/transcript.txt

Notes:

• The example reads examples/input.wav by default.
• The example does not print transcripts to stdout; it only writes examples/transcript.txt.