VisualEars FastConformer Persian ASR CoreML W4
Derivative export of Reza2kn/visualears-fastconformer-fa-full-ab.
Artifact
- Format: CoreML 4-bit k-means palettized export
- Quantization/conversion: CoreML palettize_weights, k-means, nbits=4, weight_threshold=1000000
- Runtime validation: CoreML CPU
- Validation result: 98.06% CTC argmax parity
- Size: 110 MB, 50.2% of CoreML FP16 source
Validation
Runtime parity was checked against PyTorch CTC logits on 16 calibration clips padded to 2005 mel frames. The metric is CTC argmax token agreement versus the PyTorch reference logits, not end-to-end WER.
Usage Boundary
These are fixed-frame acoustic CTC-core exports. They take precomputed log-mel features as processed_signal; they are not full raw-audio-to-text pipelines by themselves.
Notes
Best compressed verified CoreML 4-bit package. Full CoreML W4 k-means failed at 90.11% parity.
- Downloads last month
- 9
Model tree for Reza2kn/visualears-fastconformer-fa-full-ab-coreml-w4
Base model
nvidia/stt_fa_fastconformer_hybrid_large