NVIDIA
STT Ru Quartznet15x5
Model
NVIDIA
STT Ru Quartznet15x5

Speech To Text (STT) model based on QuartzNet for recognizing Russian speech.

  • 1 Version
    1.0.0rc1Selected
    03/10/2021 2:36 AM UTC67.93 MB
    Accuracy
    KeyValue
    MCV DEV SET WER (%)16.23
    Model
    KeyValue
    Inputs16000 KHZ MONO-CHANNEL AUDIO (WAV FILES)
    OutputsTRANSCRIPTION TEXT
    architecturequartznet