Wav2Vec 2.0 Base — Tibetan ASR (full fine-tune)

Fine-tuned facebook/wav2vec2-base for automatic speech recognition (ASR) on Lhasa Tibetan, released alongside:

J. Moore, S. Li and P. Lauren, "Evaluating Tibetan ASR With Segmented Word Error Rate: Beyond Character-Level Metrics," in IEEE Access, vol. 14, pp. 101790-101805, 2026, doi: 10.1109/ACCESS.2026.3709206.

This repo is one of 14 model checkpoints released with the paper, benchmarking 5 ASR architectures (Whisper Tiny/Base/Small, Wav2Vec 2.0 Base, HuBERT Base) under full fine-tuning, LoRA, and QLoRA (8-bit/4-bit) adaptation. See billingsmoore/tibetan-asr-nict-tib1-* for the full set.

Model description

Wav2Vec 2.0 is a self-supervised speech encoder pre-trained with a contrastive learning objective over masked speech frames, fine-tuned here with a CTC decoder for Tibetan transcription. This checkpoint uses full fine-tuning (all parameters updated, no quantization or adapters).

Training data

Fine-tuned on NICT-Tib1, a corpus of transcribed Lhasa Tibetan speech from 20 speakers (Soky, Gong & Li, 2022). The data was split by speaker (85/15, not by utterance) so that no speaker appears in both splits: 15,099 training utterances from 17 speakers, 1,547 test utterances from the remaining 3 speakers.

Training procedure

• Base model: facebook/wav2vec2-base
• Learning rate: 1.0e-4
• Max steps: 4000 (warmup 500)
• Effective batch size: 16 (per-device batch size 2, gradient accumulation 8)
• Decoder: CTC
• Mixed precision: fp16, with gradient checkpointing

Evaluation results

Evaluated on the 1,547-utterance NICT-Tib1 test set under five metrics: Character Error Rate (CER), Syllable Error Rate (SER), and three Segmented Word Error Rate (SWER) variants using different automatic word segmenters (BoTok-SWER, BERT-SWER, and Gem-SWER, the last computed on a 500-utterance subset for API cost reasons). All values are micro-averaged. See the paper for full bootstrap confidence intervals.

This model achieved the lowest CER of any architecture benchmarked, but its word-level SWER scores were substantially worse than Whisper Small's — a CER/SWER divergence discussed at length in the paper as a central motivating example for why CER alone is an unreliable ranking metric for Tibetan ASR.

Full model family comparison (standard fine-tuning)

How to use

from transformers import pipeline

pipe = pipeline("automatic-speech-recognition", model="billingsmoore/tibetan-asr-nict-tib1-wav2vec2-base")
result = pipe("path/to/audio.wav")
print(result["text"])

Limitations

• Trained and evaluated only on read-speech, modern Lhasa (Central) Tibetan news recordings; performance on Amdo or Kham dialects, Classical Tibetan, or spontaneous/conversational speech is unknown.
• The test set contains only 3 speakers (by design, to prevent speaker leakage), which limits how confidently results generalize across speaker, age, and prosodic variation.
• CER alone understates errors that matter at the word level for Tibetan; see the paper for why SER and SWER are necessary complements when judging output quality.

Citation

If you use this model, please cite the paper and the source NICT-Tib1 corpus:

@ARTICLE{11592371,
  author={Moore, Jacob and Li, Sheng and Lauren, Paula},
  journal={IEEE Access},
  title={Evaluating Tibetan ASR With Segmented Word Error Rate: Beyond Character-Level Metrics},
  year={2026},
  volume={14},
  number={},
  pages={101790-101805},
  keywords={Modeling;Automatic speech recognition;Error analysis;LoRa;Measurement;Ranking (statistics);Quantization (signal);Bit error rate;Standards;Training;Tibetan;automatic speech recognition;word error rate;low-resource language},
  doi={10.1109/ACCESS.2026.3709206}
}

@inproceedings{soky2022nict,
  title={Nict-tib1: A public speech corpus of lhasa dialect for benchmarking tibetan language speech recognition systems},
  author={Soky, Kak and Gong, Zhuo and Li, Sheng},
  booktitle={2022 25th Conference of the Oriental COCOSDA International Committee for the Co-ordination and Standardisation of Speech Databases and Assessment Techniques (O-COCOSDA)},
  pages={1--5},
  year={2022},
  organization={IEEE}
}

License

Released under Apache License 2.0, inherited from the base model facebook/wav2vec2-base.