TheyLoveGames releases offline Automatic Speech Recognition and Translation for Unity
- Tiny — The smallest ASR model. Fastest execution, lowest resource usage.
- Base — Larger than Tiny. Better accuracy.
- Small — Larger than Base. Significantly better accuracy.
- Medium — The largest model in this collection. Highest accuracy.
As the model sizes increase, they become more accurate but require more memory and processing power.
Platform Support: These packages are designed for Windows 64-bit (x86_64) using OnnxRuntime with DirectML hardware acceleration.
Whisper Base EN
Adds offline English speech-to-text to your Unity C# projects. This Base model offers a good balance of speed and accuracy.
Includes C# example scenes:
- Convert .WAV audio files to text
- Convert Microphone input to text in real-time
Whisper Small EN
Adds offline English speech-to-text to your Unity C# projects. This Small model provides better accuracy than Base for clearer transcription.
Includes C# example scenes:
- Convert .WAV audio files to text
- Convert Microphone input to text in real-time
Whisper Medium EN
Adds offline English speech-to-text to your Unity C# projects. This Medium model offers the highest English accuracy in this collection.
Includes C# example scenes:
- Convert .WAV audio files to text
- Convert Microphone input to text in real-time
Whisper Base Multi
Adds offline multilingual speech-to-text to your Unity C# projects. Includes the Qwen model to translate recognized text between languages. This Base model balances performance and quality.
Includes C# example scenes:
- Convert .WAV audio files to translated text
- Convert Microphone input to translated text in real-time
Whisper Small Multi
Adds offline multilingual speech-to-text to your Unity C# projects. Includes the Qwen model to translate recognized text between languages. This Small model improves transcription accuracy over Base.
Includes C# example scenes:
- Convert .WAV audio files to translated text
- Convert Microphone input to translated text in real-time
Whisper Medium Multi
Adds offline multilingual speech-to-text to your Unity C# projects. Includes the Qwen model to translate recognized text between languages. This Medium model provides the highest quality transcription and translation.
Includes C# example scenes:
- Convert .WAV audio files to translated text
- Convert Microphone input to translated text in real-time
Machine Learning packages for Unity Pending Review
Active machine learning integration using ONNX Runtime + DirectML (Windows x64, Unity Editor and Standalone, fully offline).
Audio event classification
- ml.onnxruntime.directml.yamnet.unity — (Pending Review) YAMNet package for classifying 500+ everyday sounds from clips or microphone input, with top-K labels and scores.
Whisper English speech-to-text
- ml.onnxruntime.directml.whisper.tiny.en.unity — (Pending Review) Lightweight tiny.en model for fast English transcription and dictation.
Whisper multilingual speech-to-text + translation
- ml.onnxruntime.directml.whisper.tiny.multi.unity — (Pending Review) Tiny multilingual ASR with `TranslateText(...)` powered by Qwen 2.5.
TheyLoveGames creates speech proxy
With Chrome Speech Proxy, both speech detection and synthesis can operate on non-WebGL platforms.
WebGL Speech Detection converts speech to text in real-time.
- Web Demo: Speech Dictation
- Web Demo: Speech Commands
- API: Documentation
WebGL Speech Synthesis converts text to speech in real-time.
- Web Demo: Speech Synthesis
TheyLoveGames launches two new products in the Unity Asset Store
WebGL Speech Detection converts speech to text in real-time.
- Web Demo: Speech Dictation
- Web Demo: Speech Commands
- API: Documentation
WebGL Speech Synthesis converts text to speech in real-time.
- Web Demo: Speech Synthesis
- API: Documentation
TheyLoveGames releases automation into the Unity Asset Store
Setup for Fuse CC makes Unity setup quick and easy for animation packs from Mixamo.