Modelpublic

vitthalbhandari/mms-1b-all-aft-mid-led

sourceHugging Facecc-by-nc-4.0updated 7mo agoView on Hugging Face

0likes17downloads

Model Card

MMS Adapter Fine-tuned for Lendu

This model is a fine-tuned version of facebook/mms-1b-all on the Mozilla Common Voice Spontaneous Speech dataset for Lendu (led).

Training

—Base model: facebook/mms-1b-all
—Fine-tuning method: Adapter layers
—Dataset: Mozilla Common Voice Spontaneous Speech

Usage

python

from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import torch

processor = Wav2Vec2Processor.from_pretrained("vitthalbhandari/mms-1b-all-aft-mid-led")
model = Wav2Vec2ForCTC.from_pretrained("vitthalbhandari/mms-1b-all-aft-mid-led")

# Load adapter
model.load_adapter("led")

# Transcribe audio
inputs = processor(audio_array, sampling_rate=16000, return_tensors="pt")
with torch.no_grad():
    logits = model(**inputs).logits
predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)