HoloSTT

Overview

HoloSTT is a modern, thread-safe speech recognition and input manager for Python applications. It unifies active speech, ambient listening, and keyboard input into a single, production-ready interface for any AI-driven project.

Highlights:

• Multi-modal input: Seamlessly combine voice (active/ambient) and keyboard input.
• No vendor lock-in: Works with any skill set, AI stack, or backend logic.
• Advanced audio handling: Adaptive noise management, energy thresholds, platform-aware volume support.
• Thread-safe singleton: Designed for multi-threaded and interactive desktop, assistant, and automation apps.
• Real-time text processing: Built-in cleaning, comparison, and input filtering utilities.

Why HoloSTT?

Typical speech recognition modules often come with limitations:

• Limited to just microphone input, or just keyboard.
• Lack of fallback modes, or forced use of a single speech engine.
• Not suitable for multi-threaded apps, or lacking input state management.

HoloSTT solves these problems by:

• Offering both active (“push-to-talk”) and ambient (always-on) listening, with instant keyboard fallback.
• Providing a centralized, extensible interface for all input and recognition events.
• Supporting robust error handling, state tracking, and dynamic configuration—ready for modern AI workflows.

Key Features

• Flexible Audio Capture: Toggle between active/ambient listening or switch to keyboard input instantly.

• Dynamic Noise and Volume Handling: Automatically adapts to background noise and adjusts thresholds for clear, accurate recognition.

• Centralized State & Properties: Track commands, input modes, timeouts, and all key settings in one place.

• Customizable Input Processing: Clean and filter input text, apply replacements, compare similarity, and trigger actions.

• Robust Error Handling: Handles microphone, audio, and network exceptions gracefully.

• Production-Ready: Built for real-world, scalable AI systems.

How It Works

1. Call the HoloSTT manager in your application.
2. Select input mode: active voice, ambient voice, or keyboard.
3. Process and clean recognized text automatically.
4. Use output directly for commands, automations, or AI skills.

FAQ

Q: Does HoloSTT require a specific folder or class naming? A: No. Organize your project and input logic as you see fit.

Q: Can I use my own text cleaning or filtering? A: Yes. HoloSTT exposes all processing utilities and is easy to extend.

Q: Is it production-ready and thread-safe? A: Yes. The singleton implementation ensures safe multi-threaded operation.

Code Examples

You can find code examples on my GitHub repository.

License

This project is licensed under the Apache License, Version 2.0. Copyright 2025 Tristan McBride Sr.

Authors

• Tristan McBride Sr.
• Sybil