Automatic Speech Recognition
NeMo
Safetensors
PyTorch
fastconformer
automatic-speech-translation
speech
audio
Transformer
FastConformer
Conformer
NeMo
hf-asr-leaderboard
Eval Results (legacy)
Eval Results
Instructions to use CollectionStudio/canary-1b-flash with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- NeMo
How to use CollectionStudio/canary-1b-flash with NeMo:
import nemo.collections.asr as nemo_asr asr_model = nemo_asr.models.ASRModel.from_pretrained("CollectionStudio/canary-1b-flash") transcriptions = asr_model.transcribe(["file.wav"]) - Notebooks
- Google Colab
- Kaggle
File size: 863 Bytes
96fc16b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 | {
"activation_dropout": 0.1,
"activation_function": "silu",
"architectures": [
"FastConformerModel"
],
"attention_dropout": 0.1,
"bos_token_id": 1,
"conv_kernel_size": 9,
"d_model": 1024,
"dropout": 0.1,
"dropout_emb": 0.0,
"encoder_attention_heads": 8,
"encoder_ffn_dim": 4096,
"encoder_layerdrop": 0.1,
"encoder_layers": 32,
"eos_token_id": 2,
"initializer_range": 0.02,
"model_type": "fastconformer",
"nemo_decoder_type": "none",
"nemo_encoder_type": "nemo.collections.asr.modules.ConformerEncoder",
"nemo_model_name": "nvidia/canary-1b-flash",
"nemo_model_type": "canary",
"num_mel_bins": 128,
"pad_token_id": 0,
"subsampling_conv_channels": 256,
"subsampling_factor": 8,
"torch_dtype": "float32",
"transformers_version": "4.53.0.dev0",
"use_bias": true,
"vocab_size": 1024,
"xscaling": false
}
|