Parseh aligner hi v1: ONNX int8 from ai4bharat/indicwav2vec-hindi@e746fa3e8de3f6629b9ded29d9c1c565c566f3db
354cfa6 verified | license: apache-2.0 | |
| language: hi | |
| library_name: onnx | |
| tags: | |
| - onnx | |
| - wav2vec2 | |
| - ctc | |
| - forced-alignment | |
| - quantized | |
| - parseh | |
| base_model: ai4bharat/indicwav2vec-hindi | |
| base_model_relation: quantized | |
| # Parseh Hindi forced-alignment network | |
| This CTC network aligns known Hindi caption text to 16 kHz mono audio in Parseh. It is not a general-purpose speech-recognition model. Parseh runs it locally with NumPy and ONNX Runtime, then uses CTC Viterbi alignment for word spans or character spans. | |
| ## Provenance and conversion | |
| Derived from [ai4bharat/indicwav2vec-hindi](https://huggingface.co/ai4bharat/indicwav2vec-hindi) at `e746fa3e8de3f6629b9ded29d9c1c565c566f3db`, based on `AI4Bharat IndicWav2Vec (exact upstream base revision not stated in public Hub metadata)`. It was converted to ONNX opset 17 and dynamically quantized to int8 (`QUInt8`, per-channel `MatMul` weights); no retraining was performed. Exact source and output hashes are in `meta.json` and `SHA256SUMS`. | |
| ## Verification | |
| - ONNX Runtime CPU versions: 1.23.2, 1.30.0; dynamic 0.5 s and 60 s inputs passed. | |
| - fp32 ONNX/PyTorch log-softmax max absolute difference: `6.67572021484375e-05`; frame-argmax agreement: `1.0`. | |
| - int8/fp32 frame-argmax agreement: `1.0`; generated-speech span-start delta median/p95/max: `0.0` / `0.0` / `0.0` ms. | |
| - Generated 30-second CPU alignment cost: `4.991424` s; peak RSS: `2877698048` bytes. | |
| `espeak-ng` generated audio was used only for pipeline monotonicity, not a real-speech accuracy claim. Unknown characters use a wildcard score to preserve a monotonic path. | |
| ## Limitations | |
| Input must be 16 kHz mono audio and known caption text. Noise, accents, text errors, vocabulary gaps, and quantization can affect timestamps. No accuracy guarantee is made. | |
| ## Licence and attribution | |
| The source tag is `apache-2.0`. `LICENSE` and `NOTICE` retain source, base-model, dataset attribution, and changes. No endorsement by original authors, Meta, Mozilla, or Parseh is implied. | |
| ## Citation from the source card | |
| No separate citation block was provided by the source card. | |
| ## Contact | |
| Contact: [development@parseh.io](mailto:development@parseh.io). | |