Omni Asr Ctc 300m v2 Mizonal3 E5 Lus v2026.06
> Disclaimer / Notice: Details for these are in Peer Review and publications of the paper will be made available soon for more details.
Source du modèle
- Type
- Other
- Taille du fichier
- 1,21 GB
- Modèle de base
- facebook/omniASR-CTC-300M
- Version
- vSdp 00
- Mis à jour
- 19 oct. 2026
- Source
- Hugging Face
Description de la source
Mizo Automatic Speech Recognition (ASR) Models v3.0
Disclaimer / Notice: Details for these are in Peer Review and publications of the paper will be made available soon for more details.
omni-asr-ctc-300m-v2-mizonal3-E5-lus-v2026.06
This model is a fine-tuned version of facebook/omniASR-CTC-300M on the MiZonal v3.0 dataset.
It achieves the following results on the evaluation set:
- Wer: 15.6572
- Cer: 2.7682
- Real Time Factor: 0.0060
Quick Inference
# Google Colab Ready Quick Inference
# 1. Install dependencies:
!pip install huggingface_hub omnilingual-asr torchaudio
# 2. Download the model checkpoint directly from Hugging Face:
import torch
from huggingface_hub import hf_hub_download
model_path = hf_hub_download(
repo_id="andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E5-lus-v2026.06",
filename="sdp_00.pt"
)
# 3. Load the base model pipeline and inject fine-tuned weights:
from omnilingual_asr.models.inference.pipeline import ASRInferencePipeline
print("Loading base model pipeline...")
pipeline = ASRInferencePipeline(model_card="omniASR_CTC_300M_v2")
print("Loading fine-tuned checkpoint weights...")
checkpoint = torch.load(model_path, map_location="cpu")
state_dict = checkpoint['model'] if 'model' in checkpoint else checkpoint
pipeline.model.load_state_dict(state_dict, strict=False)
# 4. Run transcription (make sure audio files are present in the specified paths):
audio_files = ["/content/segment_003.wav", "/content/segment_108.wav"]
lang = ["lus_Latn", "lus_Latn"]
transcriptions = pipeline.transcribe(audio_files, lang=lang, batch_size=2)
print("Transcription 0:", transcriptions[0])
print("Transcription 1:", transcriptions[1])
Model description
Experiment Configurations
This repository is part of a series of experiments. The different configurations are:
- E1 (Baseline): Standard training configuration.
- E2 (Noise): Training with background noise augmentation.
- E3 (Speed): Training with speed perturbation augmentation.
- E4 (SpecAug): Training with SpecAugment (time and frequency masking).
- E5 (Combined): Training with a combination of all augmentations.
All Models in this Family
| Experiment | Hugging Face Repository |
|---|---|
| E1 (Baseline) | andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E1-lus-v2026.06 |
| E2 (Noise) | andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E2-lus-v2026.06 |
| E3 (Speed) | andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E3-lus-v2026.06 |
| E4 (SpecAug) | andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E4-lus-v2026.06 |
| E5 (Combined) | andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E5-lus-v2026.06 |
Training hyperparameters
Training hyperparameters not available in bin format.
Training results
| step | epoch | train_loss | eval_loss | eval_wer | eval_cer | learning_rate | grad_norm |
|---|---|---|---|---|---|---|---|
| 100 | 2.0000 | 109.7945 | 369.5147 | ||||
| 200 | 4.0000 | 77.9571 | 136.2045 | ||||
| 300 | 6.0000 | 66.0815 | 158.5168 | ||||
| 400 | 8.0000 | 59.6202 | 104.5082 | ||||
| 500 | 10.0000 | 55.9902 | 107.9677 | ||||
| 600 | 12.0000 | 52.4551 | 106.8869 | ||||
| 700 | 13.0000 | 50.7370 | 124.5138 | ||||
| 800 | 15.0000 | 49.1970 | 95.8475 | ||||
| 900 | 17.0000 | 45.4391 | 80.7042 | ||||
| 1000 | 19.0000 | 42.8760 | 78.0457 | ||||
| 1100 | 21.0000 | 41.4012 | 36.2955 | 33.5418 | 13.3126 | 93.1231 | |
| 1200 | 23.0000 | 40.1431 | 72.7741 | ||||
| 1300 | 25.0000 | 39.0496 | 91.1769 | ||||
| 1400 | 26.0000 | 38.7845 | 68.6127 | ||||
| 1500 | 28.0000 | 37.8364 | 74.0615 | ||||
| 1600 | 30.0000 | 36.4003 | 75.8741 | ||||
| 1700 | 32.0000 | 35.1302 | 61.2816 | ||||
| 1800 | 34.0000 | 34.5149 | 71.4384 | ||||
| 1900 | 36.0000 | 33.7934 | 64.8579 | ||||
| 2000 | 38.0000 | 33.6305 | 70.0117 | ||||
| 2100 | 39.0000 | 34.9433 | 63.6476 | ||||
| 2200 | 41.0000 | 32.7293 | 34.9903 | 29.6814 | 12.2271 | 70.9184 | |
| 2300 | 43.0000 | 32.0397 | 67.0038 | ||||
| 2400 | 45.0000 | 31.1626 | 59.9783 | ||||
| 2500 | 47.0000 | 31.1168 | 61.1752 | ||||
| 2600 | 49.0000 | 31.1257 | 74.4588 | ||||
| 2700 | 50.0000 | 30.3371 | 56.0547 | ||||
| 2800 | 52.0000 | 32.3433 | 80.0529 | ||||
| 2900 | 54.0000 | 30.0935 | 56.1438 | ||||
| 3000 | 56.0000 | 29.6080 | 60.8715 | ||||
| 3100 | 58.0000 | 29.2215 | 73.8782 | ||||
| 3200 | 60.0000 | 29.2573 | 62.4088 | ||||
| 3300 | 62.0000 | 29.2438 | 35.7938 | 28.9582 | 11.2497 | 61.0301 | |
| 3400 | 63.0000 | 28.9241 | 61.3714 | ||||
| 3500 | 65.0000 | 29.4754 | 61.4402 | ||||
| 3600 | 67.0000 | 28.2900 | 56.5523 | ||||
| 3700 | 69.0000 | 27.2281 | 54.8388 | ||||
| 3800 | 71.0000 | 27.6014 | 55.5035 | ||||
| 3900 | 73.0000 | 27.4812 | 75.8993 | ||||
| 4000 | 75.0000 | 27.3107 | 64.8298 | ||||
| 4100 | 76.0000 | 28.2458 | 59.6157 | ||||
| 4200 | 78.0000 | 27.7007 | 56.1580 | ||||
| 4300 | 80.0000 | 26.8254 | 59.5118 | ||||
| 4400 | 82.0000 | 26.2679 | 39.1191 | 28.4793 | 11.8244 | 47.1478 | |
| 4500 | 84.0000 | 26.3219 | 55.5331 | ||||
| 4600 | 86.0000 | 26.7943 | 58.9898 | ||||
| 4700 | 88.0000 | 26.3254 | 55.5184 | ||||
| 4800 | 89.0000 | 27.3416 | 56.8123 | ||||
| 4900 | 91.0000 | 26.2978 | 52.2135 | ||||
| 5000 | 93.0000 | 25.5715 | 52.2345 | ||||
| 5100 | 95.0000 | 25.3241 | 56.0211 | ||||
| 5200 | 97.0000 | 25.2917 | 61.0902 | ||||
| 5300 | 99.0000 | 25.9439 | 72.8719 | ||||
| 5400 | 100.0000 | 25.5041 | 64.9717 | ||||
| 5500 | 102.0000 | 26.2498 | 36.8365 | 27.6583 | 10.7282 | 54.7318 | |
| 5600 | 104.0000 | 25.2009 | 56.9957 | ||||
| 5700 | 106.0000 | 24.6477 | 64.8322 | ||||
| 5800 | 108.0000 | 24.5322 | 61.3378 | ||||
| 5900 | 110.0000 | 24.8438 | 65.0565 | ||||
| 6000 | 112.0000 | 24.7003 | 62.3438 | ||||
| 6100 | 113.0000 | 25.4064 | 69.4132 | ||||
| 6200 | 115.0000 | 25.4864 | 66.2506 | ||||
| 6300 | 117.0000 | 24.5183 | 63.7476 | ||||
| 6400 | 119.0000 | 24.3968 | 59.3504 | ||||
| 6500 | 121.0000 | 24.2427 | 53.2391 | ||||
| 6600 | 123.0000 | 24.1187 | 38.1019 | 26.9156 | 10.8949 | 60.7356 | |
| 6700 | 125.0000 | 23.9335 | 58.0522 | ||||
| 6800 | 126.0000 | 25.3275 | 59.6575 | ||||
| 6900 | 128.0000 | 24.6366 | 63.7991 | ||||
| 7000 | 130.0000 | 24.1842 | 58.9771 | ||||
| 7100 | 132.0000 | 23.5456 | 67.6624 | ||||
| 7200 | 134.0000 | 23.9943 | 66.7438 | ||||
| 7300 | 136.0000 | 23.7824 | 61.4209 | ||||
| 7400 | 138.0000 | 24.1489 | 71.5598 | ||||
| 7500 | 139.0000 | 24.8412 | 57.2182 | ||||
| 7600 | 141.0000 | 24.3281 | 67.8653 | ||||
| 7700 | 143.0000 | 23.9259 | 37.8958 | 26.2314 | 10.5401 | 68.4402 | |
| 7800 | 145.0000 | 23.5687 | 69.3179 | ||||
| 7900 | 147.0000 | 23.7210 | 62.9377 | ||||
| 8000 | 149.0000 | 23.9842 | 70.7760 | ||||
| 8100 | 150.0000 | 23.9772 | 66.2701 | ||||
| 8200 | 152.0000 | 24.9833 | 65.0699 | ||||
| 8300 | 154.0000 | 24.3973 | 65.3704 | ||||
| 8400 | 156.0000 | 23.5754 | 64.9838 | ||||
| 8500 | 158.0000 | 23.5062 | 58.6743 | ||||
| 8600 | 160.0000 | 23.7330 | 76.8839 | ||||
| 8700 | 162.0000 | 23.9548 | 77.5334 | ||||
| 8800 | 163.0000 | 24.2769 | 37.9807 | 26.9253 | 10.6430 | 67.9240 |
Sources
1 source
Sources
1 sourceVérifié 19 oct.
Artefacts du modèle
1 artefact
Artefacts du modèle
1 artefactExtraits de sources
2 extraits
Extraits de sources
2 extraits--- language: - lus license: apache-2.0 pipeline_tag: automatic-speech-recognition base_model: facebook/omniASR-CTC-300M tags: - generated_from_trainer datasets: - andrewbawitlung/MiZonal-v3.0 metrics: - wer - cer model-index: - name: omni-asr-ctc-300m-v2-mizonal3-E5-lus-v2026.06 results: - task: name: Automatic Speech Recognition type: automatic-speech-recognition dataset: name: MiZonal v3.0 type: andrewbawitlung/MiZonal-v3.0 config: default split: test metrics: - name: Wer type: wer value: 15.6572 - name: Cer type: cer value: 2.7682 - name: Real Time Factor type: rtf value: 0.0060 ---  > **Disclaimer / Notice:** Details for these are in Peer Review and publications of the paper will be made available soon for more details. # omni-asr-ctc-300m-v2-mizonal3-E5-lus-v2026.06 This model is a fine-tuned version of [facebook/omniASR-CTC-300M](https://huggingface.co/facebook/omniASR-CTC-300M) on the **MiZonal v3.0** dataset. It achieves the following results on the evaluation set: - Wer: 15.6572 - Cer: 2.7682 - Real Time Factor: 0.0060 ## Quick Inference ```python # Google Colab Ready Quick Inference # 1. Install dependencies: !pip install huggingface_hub omnilingual-asr torchaudio # 2. Download the model checkpoint directly from Hugging Face: import torch from huggingface_hub import hf_hub_download model_path = hf_hub_download( repo_id="andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E5-lus-v2026.06", filename="sdp_00.pt" ) # 3. Load the base model pipeline and inject fine-tuned weights: from omnilingual_asr.models.inference.pipeline import ASRInferencePipeline print("Loading base model pipeline...") pipeline = ASRInferencePipeline(model_card="omniASR_CTC_300M_v2") print("Loading fine-tuned checkpoint weights...") checkpoint = torch.load(model_path, map_location="cpu") state_dict = checkpoint['model'] if 'model' in checkpoint else checkpoint pipeline.model.load_state_dict(state_dict, strict=False) # 4. Run transcription (make sure audio files are present in the specified paths): audio_files = ["/content/segment_003.wav", "/content/segment_108.wav"] lang = ["lus_Latn", "lus_Latn"] transcriptions = pipeline.transcribe(audio_files, lang=lang, batch_size=2) print("Transcription 0:", transcriptions[0]) print("Transcription 1:", transcriptions[1]) ``` ## Model descripti...
Source context: 0 downloads · 0 likes · Pipeline automatic-speech-recognition · Repo andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E5-lus-v2026.06
