Parameter	Value
Base model	`nvidia/parakeet-tdt-0.6b-v2`
Dataset	`asr-nigerian-pidgin/nigerian-pidgin-1.0`
Train samples	2,680
Val samples	666
Epochs trained	5
Batch size (effective)	32
Learning rate	5e-05
Encoder frozen	True
Best val WER	68.52%

Usage

python

import nemo.collections.asr as nemo_asr

model = nemo_asr.models.ASRModel.restore_from('parakeet-tdt-0.6b-naija-pidgin.nemo')
transcriptions = model.transcribe(['path/to/audio.wav'])
print(transcriptions)

EYEDOL/parakeet-tdt-0.6b-naija-pidgin

Parakeet-TDT-0.6B-v2 fine-tuned on Nigerian Pidgin (Naija)

Training

Usage

EYEDOL/
parakeet-tdt-0.6b-naija-pidgin