Summary

FUTO Voice Input is an Android application providing speech-to-text capabilities, designed for users seeking to integrate voice input with third-party keyboards or applications leveraging standard Android speech-to-text APIs. It offers an alternative to proprietary, built-in solutions, enhancing flexibility for users of various keyboards.

How It Works

The core of FUTO Voice Input relies on the OpenAI Whisper model for speech recognition. It integrates with Android via two primary mechanisms: the android.speech.action.RECOGNIZE_SPEECH implicit intent for floating window UIs and the IME voice subtype mode for on-screen keyboard integration. This design prioritizes compatibility with keyboards supporting these standard Android speech interfaces.

Quick Start & Requirements

• Build: ./gradlew assembleStandaloneRelease using Gradle.
• Dependencies: Android Studio or Gradle. Prebuilt binaries are available in libs/ for accelerated builds.
• Resources: Official download and support information are available via the FUTO Voice Input page and public issue tracker.

Highlighted Details

• Language Support: Utilizes OpenAI Whisper, with UI options limited to languages having over 1,000 training hours (e.g., English, Spanish, French, German, Chinese, Japanese, Korean).
• Keyboard Compatibility: Supports HeliBoard, FlorisBoard, AnySoftKeyboard, Unexpected Keyboard, AOSP Keyboard, Grammarly, and SwiftKey. Notably incompatible with Gboard and Samsung Keyboard due to their hardcoded voice input restrictions.
• Build Flavors: Offers dev, playStore, standalone, and fDroid variants, differing in features like Play Store billing and auto-update capabilities.

Maintenance & Community

Development focus has largely shifted to the FUTO Keyboard app, which includes built-in voice input. FUTO Voice Input remains available as a standalone option. Feedback and issues are welcomed on the public issue tracker.

Licensing & Compatibility

Licensed under the "FUTO Source First License 1.0". Specific compatibility notes for commercial use are not detailed, but the license suggests an open-source orientation.

Limitations & Caveats

Currently excludes support for the SpeechRecognizer API. Chinese language input exhibits known behavioral quirks. The project's primary development effort has transitioned to the FUTO Keyboard.