CreativeSkipper/MynaLuna-ASR

MynaLuna ASR

TThis repository provides an INT8-quantized ONNX model distributio of NVIDIA NeMo Parakeet TDT 0.6B v3 for use with the MynaLuna application.

Repository purpose

This repository is maintained to provide a stable download source for the MynaLuna application.

It helps ensure that users can continue downloading the required model even if the original distribution source changes or becomes unavailable.

Model details

Property	Value
Purpose	Automatic Speech Recognition (ASR)
Base model	NVIDIA NeMo Parakeet TDT 0.6B v3
Architecture	FastConformer-TDT (Transducer)
Quantization	INT8
Format	ONNX
Runtime	sherpa-onnx
Encoder size	~622 MB

Files

encoder.int8.onnx   — Acoustic encoder (FastConformer)
decoder.int8.onnx   — Prediction network
joiner.int8.onnx    — Joint network
tokens.txt          — BPE vocabulary (8,192 tokens)
test_wavs/          — Reference audio samples

Attribution

Base model:

https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3

ONNX export:

https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8.tar.bz2

License

This repository redistributes a derivative of the original model.

The original NVIDIA model is licensed under CC BY 4.0.

Please refer to the original repository for licensing details:

https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3