Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

fix bug

游雁f3dfbeed2d89
diff --git a/examples/industrial_data_pretraining/llm_asr/demo_speech2text.sh b/examples/industrial_data_pretraining/llm_asr/demo_speech2text.sh
index 81f211fa8..6b5f228da 100644
--- a/examples/industrial_data_pretraining/llm_asr/demo_speech2text.sh
+++ b/examples/industrial_data_pretraining/llm_asr/demo_speech2text.sh
@@ -13,8 +13,8 @@ for data_set in "librispeech_test_clean_speech2text.jsonl" "librispeech_test_oth
jsonl=${jsonl_dir}/${data_set}
output_dir=${out_dir}/${data_set}
mkdir -p ${output_dir}
- pred_file=${output_dir}/${data_set}/1best_recog/text_tn
- ref_file=${output_dir}/${data_set}/1best_recog/label
+ pred_file=${output_dir}/1best_recog/text_tn
+ ref_file=${output_dir}/1best_recog/label
python ./demo_speech2text.py ${ckpt_dir} ${ckpt_id} ${jsonl} ${output_dir} ${device}
@@ -27,12 +27,12 @@ for data_set in "aishell1_test_speech2text.jsonl" "aishell2_ios_test_speech2text
jsonl=${jsonl_dir}/${data_set}
output_dir=${out_dir}/${data_set}
mkdir -p ${output_dir}
- pred_file=${output_dir}/${data_set}/1best_recog/text_tn
- ref_file=${output_dir}/${data_set}/1best_recog/label
+ pred_file=${output_dir}/1best_recog/text_tn
+ ref_file=${output_dir}/1best_recog/label
- python ./demo_speech2text.py ${ckpt_dir} ${ckpt_id} ${jsonl} ${output_dir}
+ 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
+ python /mnt/workspace/zhifu.gzf/codebase/FunASR/funasr/metrics/wer.py ++ref_file=${ref_file} ++hyp_file=${pred_file} ++cer_file=${pred_file}.cer ++cn_postprocess=false
done
@@ -40,12 +40,12 @@ for data_set in "s2tt_en2zh.v20240605.test.jsonl"; do
jsonl=${jsonl_dir}/${data_set}
output_dir=${out_dir}/${data_set}
mkdir -p ${output_dir}
- pred_file=${output_dir}/${data_set}/1best_recog/text_tn
- ref_file=${output_dir}/${data_set}/1best_recog/label
+ pred_file=${output_dir}/1best_recog/text_tn
+ ref_file=${output_dir}/1best_recog/label
- python ./demo_speech2text.py ${ckpt_dir} ${ckpt_id} ${jsonl} ${output_dir}
+ 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
+ python /mnt/workspace/zhifu.gzf/codebase/FunASR/funasr/metrics/wer.py ++ref_file=${ref_file} ++hyp_file=${pred_file} ++cer_file=${pred_file}.cer ++cn_postprocess=false
done
@@ -53,11 +53,11 @@ for data_set in "s2tt_zh2en.v20240605.test.jsonl"; do
jsonl=${jsonl_dir}/${data_set}
output_dir=${out_dir}/${data_set}
mkdir -p ${output_dir}
- pred_file=${output_dir}/${data_set}/1best_recog/text_tn
- ref_file=${output_dir}/${data_set}/1best_recog/label
+ pred_file=${output_dir}/1best_recog/text_tn
+ ref_file=${output_dir}/1best_recog/label
- python ./demo_speech2text.py ${ckpt_dir} ${ckpt_id} ${jsonl} ${output_dir}
+ 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
+ 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
\ No newline at end of file