inference.sh

#!/usr/bin/env bash

# export PYTHONPATH=./../../../:$PYTHONPATH

python zipformer_inference/inference_model.py \
    --ckpt-path 327M-uni-v2-batch-mix-0.3-p-n0.5-400k/iter-400000-avg-4.pt \
    --output-downsampling-factor 1 \
    --downsampling-factor 1,2,4,8,4,2,1 \
    --num-encoder-layers 1,2,2,3,1,1,1 \
    --feedforward-dim 3072,3072,3072,3072,3072,3072,3072 \
    --encoder-dim 1024,1024,1024,1024,1024,1024,1024 \
    --encoder-unmasked-dim 512,512,512,512,512,512,512 \
    --cnn-module-kernel 31,31,15,15,15,31,31 \
    --num-heads 8,8,8,8,8,8,8 \
    --audio 5338-24640-0000.flac