Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
test
语帆5f91acae0d8b
diff --git a/examples/industrial_data_pretraining/lcbnet/demo_nj.sh b/examples/industrial_data_pretraining/lcbnet/demo_nj.shdeleted file mode 100755index d9f42a033..000000000--- a/examples/industrial_data_pretraining/lcbnet/demo_nj.sh+++ /dev/null@@ -1,72 +0,0 @@-file_dir="/nfs/yufan.yf/workspace/github/FunASR/examples/industrial_data_pretraining/lcbnet/exp/speech_lcbnet_contextual_asr-en-16k-bpe-vocab5002-pytorch"-CUDA_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"-inference_device="cuda"--if [ ${inference_device} == "cuda" ]; then- nj=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')-else- inference_batch_size=1- CUDA_VISIBLE_DEVICES=""- for JOB in $(seq ${nj}); do- CUDA_VISIBLE_DEVICES=$CUDA_VISIBLE_DEVICES"-1,"- done-fi--inference_dir="outputs/test"-_logdir="${inference_dir}/logdir"-echo "inference_dir: ${inference_dir}"--# mkdir -p "${_logdir}"-# key_file1=${file_dir}/wav.scp-# key_file2=${file_dir}/ocr.txt-# split_scps1=-# split_scps2=-# for JOB in $(seq "${nj}"); do-# split_scps1+=" ${_logdir}/wav.${JOB}.scp"-# split_scps2+=" ${_logdir}/ocr.${JOB}.txt"-# done-# utils/split_scp.pl "${key_file1}" ${split_scps1}-# utils/split_scp.pl "${key_file2}" ${split_scps2}--# gpuid_list_array=(${CUDA_VISIBLE_DEVICES//,/ })-# for JOB in $(seq ${nj}); do-# {-# id=$((JOB-1))-# gpuid=${gpuid_list_array[$id]}--# export CUDA_VISIBLE_DEVICES=${gpuid}--# python -m funasr.bin.inference \-# --config-path=${file_dir} \-# --config-name="config.yaml" \-# ++init_param=${file_dir}/model.pb \-# ++tokenizer_conf.token_list=${file_dir}/tokens.txt \-# ++input=[${_logdir}/wav.${JOB}.scp,${_logdir}/ocr.${JOB}.txt] \-# +data_type='["kaldi_ark", "text"]' \-# ++tokenizer_conf.bpemodel=${file_dir}/bpe.model \-# ++output_dir="${inference_dir}/${JOB}" \-# ++device="${inference_device}" \-# ++ncpu=1 \-# ++disable_log=true &> ${_logdir}/log.${JOB}.txt--# }&-# done-# wait---#mkdir -p ${inference_dir}/1best_recog-for f in token; do- if [ -f "${inference_dir}/${JOB}/1best_recog/${f}" ]; then- for JOB in $(seq "${nj}"); do- cat "${inference_dir}/${JOB}/1best_recog/${f}"- done | sort -k1 >"${inference_dir}/1best_recog/${f}"- fi-done--echo "Computing WER ..."-echo "Computing WER ..."-#python utils/postprocess_text_zh.py ${inference_dir}/1best_recog/text ${inference_dir}/1best_recog/text.proc--#cp ${data_dir}/text ${inference_dir}/1best_recog/text.ref-#python utils/compute_wer.py ${inference_dir}/1best_recog/text.ref ${inference_dir}/1best_recog/text.proc ${inference_dir}/1best_recog/text.cer-#tail -n 3 ${inference_dir}/1best_recog/text.cer\ No newline at end of filediff --git a/examples/industrial_data_pretraining/lcbnet/demo_nj2.sh b/examples/industrial_data_pretraining/lcbnet/demo_nj2.shnew file mode 100644index 000000000..205c28fa3--- /dev/null+++ b/examples/industrial_data_pretraining/lcbnet/demo_nj2.sh@@ -0,0 +1,72 @@+file_dir="/nfs/yufan.yf/workspace/github/FunASR/examples/industrial_data_pretraining/lcbnet/exp/speech_lcbnet_contextual_asr-en-16k-bpe-vocab5002-pytorch"+CUDA_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"+inference_device="cuda"++if [ ${inference_device} == "cuda" ]; then+ nj=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')+ else+ inference_batch_size=1+ CUDA_VISIBLE_DEVICES=""+ for JOB in $(seq ${nj}); do+ CUDA_VISIBLE_DEVICES=$CUDA_VISIBLE_DEVICES"-1,"+ done+ fi++ inference_dir="outputs/test"+ _logdir="${inference_dir}/logdir"+ echo "inference_dir: ${inference_dir}"++ # mkdir -p "${_logdir}"+ # key_file1=${file_dir}/wav.scp+ # key_file2=${file_dir}/ocr.txt+ # split_scps1=+ # split_scps2=+ # for JOB in $(seq "${nj}"); do+ # split_scps1+=" ${_logdir}/wav.${JOB}.scp"+ # split_scps2+=" ${_logdir}/ocr.${JOB}.txt"+ # done+ # utils/split_scp.pl "${key_file1}" ${split_scps1}+ # utils/split_scp.pl "${key_file2}" ${split_scps2}++ # gpuid_list_array=(${CUDA_VISIBLE_DEVICES//,/ })+ # for JOB in $(seq ${nj}); do+ # {+ # id=$((JOB-1))+ # gpuid=${gpuid_list_array[$id]}++ # export CUDA_VISIBLE_DEVICES=${gpuid}++ # python -m funasr.bin.inference \+ # --config-path=${file_dir} \+ # --config-name="config.yaml" \+ # ++init_param=${file_dir}/model.pb \+ # ++tokenizer_conf.token_list=${file_dir}/tokens.txt \+ # ++input=[${_logdir}/wav.${JOB}.scp,${_logdir}/ocr.${JOB}.txt] \+ # +data_type='["kaldi_ark", "text"]' \+ # ++tokenizer_conf.bpemodel=${file_dir}/bpe.model \+ # ++output_dir="${inference_dir}/${JOB}" \+ # ++device="${inference_device}" \+ # ++ncpu=1 \+ # ++disable_log=true &> ${_logdir}/log.${JOB}.txt++ # }&+ # done+ # wait+++ #mkdir -p ${inference_dir}/1best_recog+ for f in token; do+ if [ -f "${inference_dir}/${JOB}/1best_recog/${f}" ]; then+ for JOB in $(seq "${nj}"); do+ cat "${inference_dir}/${JOB}/1best_recog/${f}"+ done | sort -k1 >"${inference_dir}/1best_recog/${f}"+ fi+ done++ echo "Computing WER ..."+ echo "Computing WER ..."+ #python utils/postprocess_text_zh.py ${inference_dir}/1best_recog/text ${inference_dir}/1best_recog/text.proc++ #cp ${data_dir}/text ${inference_dir}/1best_recog/text.ref+ #python utils/compute_wer.py ${inference_dir}/1best_recog/text.ref ${inference_dir}/1best_recog/text.proc ${inference_dir}/1best_recog/text.cer+ #tail -n 3 ${inference_dir}/1best_recog/text.cerdiff --git a/examples/industrial_data_pretraining/lcbnet/utils b/examples/industrial_data_pretraining/lcbnet/utilsnew file mode 120000index 000000000..be5e5a322--- /dev/null+++ b/examples/industrial_data_pretraining/lcbnet/utils@@ -0,0 +1 @@+../../aishell/paraformer/utils\ No newline at end of file