indicconformer_as_finetune_v1
Fine-tuned from ai4bharat/indicconformer_stt_as_hybrid_rnnt_large on a 87hours (40k sample) subset of ananddey/assamese-asr-dataset.
Usage
import nemo.collections.asr as nemo_asr
model = nemo_asr.models.ASRModel.from_pretrained("amn-raw/indic_conformer_ctc_rnnt_finetuned")
model.cur_decoder = "ctc"
print(model.transcribe(["sample_audio_16k_mono.wav"], batch_size=1, language_id="as")[0])
Note: this is a fine-tune of an AI4Bharat model — check the base model's license before redistributing.
- Downloads last month
- 15