theylovegames.com

What It Is

TheyLoveGames currently showcases a set of machine-learning asset packages for Unity, with core capabilities including offline automatic speech recognition, translation, and audio analysis. The products are distributed through the Unity Asset Store and are mainly aimed at developers who need to add voice interaction, real-time captions, voice commands, or sound-event recognition to Unity C# projects.

Core Capabilities and Tech Stack

Its ASR solution is based on Whisper, with both English-only and multilingual versions, and model sizes divided into Tiny, Base, Small, and Medium. The product description makes it clear that larger models deliver higher accuracy, but also require more memory and processing power. The multilingual package also includes a Qwen model for translating recognized text between languages. The Yamnet package is used for offline audio-event classification, recognizing 521 categories of everyday sounds and supporting both audio clips and microphone input.

On the platform side, these packages are designed for Windows 64-bit x86_64, using ONNX Runtime and DirectML for hardware acceleration. They cover both Unity Editor and Standalone builds, with an emphasis on being fully offline. Each package provides C# sample scenes, such as .WAV-to-text transcription, real-time microphone transcription, and real-time sound classification.

Pricing and Documentation

The captured content does not disclose pricing, licensing model, payment methods, or upgrade policy; the only confirmed point is that the products are listed on the Unity Asset Store. In terms of documentation, each package is marked as having Documentation and includes sample scenes. The older WebGL Speech Detection/Synthesis product also provides API documentation and a Web Demo. However, the full documentation content is not shown in the source text, so we can only confirm that basic documentation entry points exist; its quality cannot be assessed in depth.

Pros and Cons

The main strengths are offline operation, Unity-focused integration, coverage of ASR/translation/audio classification, and multiple model sizes that let developers balance performance against accuracy. This is especially valuable for games and interactive applications that prefer not to rely on cloud speech services. The drawbacks are that platform support is clearly concentrated on Windows x64, cross-platform information is limited, and the models can be large—for example, Medium Multi reaches 2.6 GB. Pricing and technical support details are also missing, and some packages are still Pending Review.

Best Fit and Access from China

It is best suited for Windows desktop Unity games, exhibition/interactives, educational apps, offline captioning, or voice-command projects. If a project requires mobile, WebGL, macOS, or Linux support, it should be evaluated carefully. The source text provides no basis for assessing access from China, so it is marked as unknown. If access to or payment on the Unity Asset Store is restricted, alternatives include integrating Whisper/ONNX independently or choosing domestic cloud speech services.