Small fastconformer CTC ASR model

A small fastconformer CTC ASR model for light, especially low to medium vocabulary applications.

Model Details

Model Description

  • Developed by: Vojtěch Drábek
  • Model type: Fastconformer CTC
  • Language(s) (NLP): Czech
  • License: CC-BY-4.0

Evaluation

Testing Data

Mozilla Common Voice Czech 21.0 Test Set

Results

Greedy decoding

Full model

WER: 15.0 CER: 5.2

Quantized model

WER: 15.5 CER: 5.4

Been search decoding

General language model
  • pyctcdecode
  • beam width 16
  • 5-gram, 500k words, 992 MB

WER: 11.6 CER: 5.8

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support