singscope-align

CTC forced-alignment models (INT8 dynamic quantization, ONNX) used by SingScope to refine note boundaries from sung lyrics.

• en/: quantized from facebook/wav2vec2-large-960h-lv60-self (Apache-2.0) via torch.onnx.export (opset 17) + onnxruntime quantize_dynamic
• ja/: quantized from reazon-research/japanese-wav2vec2-large-rs35kh (Apache-2.0), same procedure

These are unmodified quantized derivatives of the above Apache-2.0 models. See the original model cards for training details and intended use.