This Python SDK provides a comprehensive interface for interacting with Deepgram's AI-powered speech and language models, targeting developers building applications that require transcription, text-to-speech, and text analysis. It offers synchronous and asynchronous APIs for various audio sources and streaming capabilities, simplifying integration with Deepgram's advanced AI services.

How It Works

The SDK leverages a RESTful API for pre-recorded audio processing and a WebSocket API for real-time streaming and voice agent interactions. It abstracts the complexities of network requests, authentication, and data serialization, allowing developers to focus on their application logic. The library supports various models for transcription (e.g., nova-3) and text-to-speech (e.g., aura-2-thalia-en), with options for language, callbacks, and advanced agent configurations.

Quick Start & Requirements

• Primary install: pip install deepgram-sdk
• Requirements: Python 3.10+
• Documentation: developers.deepgram.com

Highlighted Details

• Supports both API key and bearer token authentication, with dynamic switching.
• Offers dedicated modules for managing projects, keys, members, and usage.
• Includes capabilities for converting transcriptions to WebVTT and SRT formats.
• Provides extensive support for Voice Agent configuration, integrating with models like OpenAI.

Maintenance & Community

• Follows semantic versioning for backward compatibility within major releases.
• Active community support via GitHub Discussions and Discord.
• Contribution guidelines are available for developers.

Licensing & Compatibility

• The SDK is released under the MIT License, permitting commercial use and closed-source linking.

Limitations & Caveats

• Migration guides are provided for v2 to v3 and v3.* to v4, indicating potential breaking changes between major versions.
• Running daily tests requires a DEEPGRAM_API_KEY to be set in environment variables.