Skip to content
ModelListedExternal
Download

Omni Asr Ctc 300m v2 Mizonal3 E3 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.

Sourcehuggingface.co/andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E3-lus-v2026.06/tree/299330e82a03c39301ff521f983841a061b2b21f
KindOtherBase modelfacebook/omniASR-CTC-300MVersionSdp 00Cgrade

Model source

Kind
Other
File size
1.21 GB
Base model
facebook/omniASR-CTC-300M
Version
vSdp 00
Updated
Oct 19, 2026
Source
Hugging Face

Source description

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-E3-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: 15.7161
  • Cer: 2.8604
  • Real Time Factor: 0.0064

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-E3-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

ExperimentHugging 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

stepepochtrain_losseval_losseval_wereval_cerlearning_rategrad_norm
1003.000065.5965281.3520
2005.000039.098488.8331
3008.000032.059798.2777
40010.000023.724459.9031
50013.000020.823979.3340
60015.000016.305747.8548
70018.000014.492753.2894
80020.000011.279512.589320.01564.016038.7695
90023.00009.888635.1124
100025.00008.166528.8110
110028.00007.124127.0514
120030.00005.554825.8380
130033.00005.067826.2448
140035.00004.334924.4729
150038.00003.778724.0198
160040.00003.493315.942316.89803.263918.6042
170043.00002.761118.6376
180045.00002.357517.8721
190048.00002.454018.4342
200050.00002.215415.9295
210053.00001.877118.9938
220055.00001.571012.7145
230058.00001.509215.3270
240060.00001.437422.227115.94023.111313.3519
250063.00001.293013.1228
260065.00001.229112.2031
270068.00001.222212.3354
280070.00001.290511.6585
290073.00000.995210.9727
300075.00001.009710.5416
310078.00000.910917.1557
320080.00000.736024.956715.54932.99079.6445
330083.00000.693310.1479
340085.00000.69049.1531
350088.00000.65769.4965
360090.00000.58757.7222
370093.00000.70399.1735
380095.00000.567813.2920
390098.00000.53728.1872
4000100.00000.471026.338414.54262.81516.2237
4100103.00000.50478.6809
4200105.00000.38695.7722
4300108.00000.38776.3638
4400110.00000.40786.3570
4500113.00000.669510.3843
4600115.00000.37206.5772
4700118.00000.42187.0805
4800120.00000.333927.558714.50352.80445.6652
4900123.00000.33876.7879
5000125.00000.30846.1141
5100128.00000.31117.2847
5200130.00000.35745.8786
5300133.00000.302814.1547
5400135.00000.22354.3351
5500138.00000.39685.9151
5600140.00000.277527.916214.24942.77434.8575
5700143.00000.27554.6821
5800145.00000.30465.0620
5900148.00000.28124.9860
6000150.00000.29825.3985
6100153.00000.27275.1145
6200155.00000.26424.7418
6300158.00000.33875.6535
6400160.00000.275928.494314.10282.72825.1401

Sources

1 source
Hugging FacePrimaryVersion 299330e82a03c39301ff521f983841a061b2b21f
huggingface.co/andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E3-lus-v2026.06/tree/299330e82a03c39301ff521f983841a061b2b21f

Verified Oct 19

Model artifacts

1 artifact

sdp_00.pt

pt · 1.21 GB · SHA-256 25d17b8da3c3…f98f · Hugging Face

Download file

Source excerpts

2 excerpts
ReadmeAug 3huggingface.co/andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E3-lus-v2026.06

--- 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-E3-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: 15.7161 - name: Cer type: cer value: 2.8604 - name: Real Time Factor type: rtf value: 0.0064 --- ![Mizo Automatic Speech Recognition (ASR) Models v3.0](banner.jpg) > **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-E3-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: 15.7161 - Cer: 2.8604 - Real Time Factor: 0.0064 ## 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-E3-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...

NoteAug 3huggingface.co/andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E3-lus-v2026.06

Source context: 0 downloads · 0 likes · Pipeline automatic-speech-recognition · Repo andrewbawitlung/omni-asr-ctc-300m-v2-mizonal3-E3-lus-v2026.06