You can deploy this model and start earning money today!

slyusarev/
parakeet-tdt-0.6b-v3

Name: parakeet-tdt-0.6b-v3
Availability: InStock

automatic-speech-recognition

Views

No views yet

Parakeet TDT 0.6B v3 — ONNX (int8)

nvidia/parakeet-tdt-0.6b-v3 exported to ONNX and quantized to int8. Built for Textolit, an offline dictation app — the app downloads these files automatically.

File	Size	SHA-256
`encoder-model.int8.onnx`	654 MB	`7897d54aa72b4f280c816dc59d296ee3f6436180b69088fba0c57806ea6d840d`
`decoder_joint-model.int8.onnx`	18 MB	`05892ef89093e7dabcf23b83b4123089ff00231f6556b1b67f4350eaf454b5e5`
`vocab.txt`	94 KB	`d58544679ea4bc6ac563d1f545eb7d474bd6cfa467f0a6e2c1dc1c7d37e3c35d`
`config.json`	98 B	`21565538716b1cb134a096e3df9fa3f27f0470bc09c596d9faa578dc27d945d5`

The audio preprocessor is not included: Textolit computes mel features itself.

License: CC-BY-4.0, inherited from the original. Attribution: NVIDIA; the export to ONNX and the int8 quantization named above are the only changes to their model — nothing was retrained.

slyusarev/parakeet-tdt-0.6b-v3

Parakeet TDT 0.6B v3 — ONNX (int8)

slyusarev/
parakeet-tdt-0.6b-v3