AutoArk/GPA

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

2762 stars · 186 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

GPA (General Purpose Audio) is a unified autoregressive audio-language model that performs text-to-speech, automatic speech recognition, and voice conversion within a single small model. It offers native PyTorch/Hugging Face workflows plus an ONNX runtime with CLI, FastAPI service, and browser UI deployment options.

Use cases

• transcribe speech to text with a tiny model
• synthesize speech from text with voice cloning
• convert one voice to another
• deploy speech models on edge devices with ONNX
• run a self-hosted speech API service
• fine-tune a unified audio model on custom data

When to choose

• you want one compact model covering ASR, TTS, and voice conversion instead of separate pipelines
• you need lightweight, quantized speech runtimes for edge or CPU deployment
• you want an open Apache-2.0 speech model with ONNX, FastAPI, and browser UI options

When to avoid

• you need state-of-the-art accuracy on a single specific speech task and can afford larger specialized models
• you need low-latency streaming or production-grade enterprise speech services
• you require extensive language coverage beyond what the model supports

Facets

library · maturity active

speech-recognition tts machine-learning llm-inference audio-processing speech-processing artificial-intelligence deep-learning python cross-platform cli audio-language-model voice-conversion onnx-runtime unified-model speech-synthesis edge-deployment quantization audio gpu web-server

Member repositories

For agents

markdown · JSON · MCP: product_card(name="AutoArk/GPA")