mirror of
https://github.com/modelscope/FunASR
synced 2025-09-15 14:48:36 +08:00
64 lines
2.4 KiB
Bash
64 lines
2.4 KiB
Bash
|
|
ckpt_dir="/nfs/beinian.lzr/workspace/GPT-4o/Exp/exp6/5m-8gpu/exp6_speech2text_linear_ddp_0609"
|
|
ckpt_id="model.pt.ep0.90000"
|
|
jsonl_dir="/nfs/beinian.lzr/workspace/GPT-4o/Data/Speech2Text/TestData"
|
|
out_dir="${ckpt_dir}/inference-${ckpt_id}"
|
|
mkdir -p ${out_dir}
|
|
|
|
device="cuda:0"
|
|
|
|
ckpt_id=$1
|
|
device=$2
|
|
|
|
for data_set in "librispeech_test_clean_speech2text.jsonl" "librispeech_test_other_speech2text.jsonl"; do
|
|
jsonl=${jsonl_dir}/${data_set}
|
|
output_dir=${out_dir}/${data_set}
|
|
mkdir -p ${output_dir}
|
|
pred_file=${output_dir}/1best_recog/text_tn
|
|
ref_file=${output_dir}/1best_recog/label
|
|
|
|
python ./demo_speech2text.py ${ckpt_dir} ${ckpt_id} ${jsonl} ${output_dir} ${device}
|
|
|
|
python /mnt/workspace/zhifu.gzf/codebase/FunASR/funasr/metrics/wer.py ++ref_file=${ref_file} ++hyp_file=${pred_file} ++cer_file=${pred_file}.cer ++cn_postprocess=false
|
|
|
|
done
|
|
|
|
|
|
for data_set in "aishell1_test_speech2text.jsonl" "aishell2_ios_test_speech2text.jsonl" "librispeech_test_other_speech2text.jsonl"; do
|
|
jsonl=${jsonl_dir}/${data_set}
|
|
output_dir=${out_dir}/${data_set}
|
|
mkdir -p ${output_dir}
|
|
pred_file=${output_dir}/1best_recog/text_tn
|
|
ref_file=${output_dir}/1best_recog/label
|
|
|
|
python ./demo_speech2text.py ${ckpt_dir} ${ckpt_id} ${jsonl} ${output_dir} ${device}
|
|
|
|
python /mnt/workspace/zhifu.gzf/codebase/FunASR/funasr/metrics/wer.py ++ref_file=${ref_file} ++hyp_file=${pred_file} ++cer_file=${pred_file}.cer ++cn_postprocess=true
|
|
|
|
done
|
|
|
|
for data_set in "s2tt_en2zh.v20240605.test.jsonl"; do
|
|
jsonl=${jsonl_dir}/${data_set}
|
|
output_dir=${out_dir}/${data_set}
|
|
mkdir -p ${output_dir}
|
|
pred_file=${output_dir}/1best_recog/text_tn
|
|
ref_file=${output_dir}/1best_recog/label
|
|
|
|
python ./demo_speech2text.py ${ckpt_dir} ${ckpt_id} ${jsonl} ${output_dir} ${device}
|
|
|
|
python /mnt/workspace/zhifu.gzf/codebase/FunASR/funasr/metrics/wer.py ++ref_file=${ref_file} ++hyp_file=${pred_file} ++cer_file=${pred_file}.cer ++cn_postprocess=true
|
|
|
|
done
|
|
|
|
for data_set in "s2tt_zh2en.v20240605.test.jsonl"; do
|
|
jsonl=${jsonl_dir}/${data_set}
|
|
output_dir=${out_dir}/${data_set}
|
|
mkdir -p ${output_dir}
|
|
pred_file=${output_dir}/1best_recog/text_tn
|
|
ref_file=${output_dir}/1best_recog/label
|
|
|
|
python ./demo_speech2text.py ${ckpt_dir} ${ckpt_id} ${jsonl} ${output_dir} ${device}
|
|
|
|
python /mnt/workspace/zhifu.gzf/codebase/FunASR/funasr/metrics/wer.py ++ref_file=${ref_file} ++hyp_file=${pred_file} ++cer_file=${pred_file}.cer ++cn_postprocess=false
|
|
|
|
done |