Tất cả mô hình
N

Nemotron-3.5 ASR Streaming 0.6B

bởi nvidiaCông khaiopenmdw-1.1
README.md

Nemotron-3.5 ASR Streaming 0.6B

Nemotron-3.5 ASR Streaming 0.6B is NVIDIA's multilingual, cache-aware streaming automatic-speech-recognition model. It uses a FastConformer encoder and RNN-T decoder.

Links

  • Upstream model card: https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b
  • CrispASR backend: https://github.com/CrispStrobe/CrispASR/releases/tag/v0.8.6

Model Facts

  • Task: Speech-to-Text
  • Parameters: 0.6B
  • License: OpenMDW-1.1
  • Languages: 35 languages, including Vietnamese
  • Architecture: Cache-aware FastConformer RNN-T

LA Studio Notes

LA Studio runs this model locally with the nemotron backend in CrispASR v0.8.6 or later. The app downloads GGUF artifacts from cstr/nemotron-3.5-asr-streaming-0.6b-GGUF:

  • nemotron-3.5-asr-streaming-0.6b-q4_k.gguf (recommended, 479 MB)
  • nemotron-3.5-asr-streaming-0.6b-f16.gguf (1.3 GB)

The model is streaming-native. LA Studio's current STT workflow transcribes selected or recorded audio as a completed job; live incremental transcript delivery requires a future CrispASR session-stream API for Nemotron.

Cấu hình tham số quy trình

Cấu hình các tham số tùy chỉnh này để tinh chỉnh quy trình tiếng nói và giọng nói.

Temperaturefloat

Controls randomness of speech generation. Higher values increase variety.

1
Seedint

Random seed for reproducibility. Set to -1 for random.

-1
Max Codec Stepsint

Maximum decoding steps for the neural voice audio synthesizer.

100