initial commit

FunASR/examples/industrial_data_pretraining/bicif_paraformer/export.sh

+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+# method1, inference from model hub
+export HYDRA_FULL_ERROR=1
+
+
+model="iic/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch"
+
+python -m funasr.bin.export \
+++model=${model} \
+++type="onnx" \
+++quantize=false \
+++device="cpu"
+
+# method2, inference from local path
+model="/Users/zhifu/.cache/modelscope/hub/iic/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch"
+
+python -m funasr.bin.export \
+++model=${model} \
+++type="onnx" \
+++quantize=false \
+++device="cpu"
\ No newline at end of file

FunASR/examples/industrial_data_pretraining/bicif_paraformer/finetune.sh

+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+workspace=`pwd`
+
+# method1, finetune from model hub
+
+# which gpu to train or finetune
+export CUDA_VISIBLE_DEVICES="0,1"
+gpu_num=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')
+
+# model_name from model_hub, or model_dir in local path
+
+## option 1, download model automatically
+model_name_or_model_dir="iic/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch"
+
+## option 2, download model by git
+#local_path_root=${workspace}/modelscope_models
+#mkdir -p ${local_path_root}/${model_name_or_model_dir}
+#git clone https://www.modelscope.cn/${model_name_or_model_dir}.git ${local_path_root}/${model_name_or_model_dir}
+#model_name_or_model_dir=${local_path_root}/${model_name_or_model_dir}
+
+
+# data dir, which contains: train.json, val.json
+data_dir="../../../data/list"
+
+train_data="${data_dir}/train.jsonl"
+val_data="${data_dir}/val.jsonl"
+
+# generate train.jsonl and val.jsonl from wav.scp and text.txt
+scp2jsonl \
+++scp_file_list='["../../../data/list/train_wav.scp", "../../../data/list/train_text.txt"]' \
+++data_type_list='["source", "target"]' \
+++jsonl_file_out="${train_data}"
+
+scp2jsonl \
+++scp_file_list='["../../../data/list/val_wav.scp", "../../../data/list/val_text.txt"]' \
+++data_type_list='["source", "target"]' \
+++jsonl_file_out="${val_data}"
+
+
+# exp output dir
+output_dir="./outputs"
+log_file="${output_dir}/log.txt"
+
+
+mkdir -p ${output_dir}
+echo "log_file: ${log_file}"
+
+deepspeed_config=${workspace}/../../ds_stage1.json
+
+DISTRIBUTED_ARGS="
+    --nnodes ${WORLD_SIZE:-1} \
+    --nproc_per_node $gpu_num \
+    --node_rank ${RANK:-0} \
+    --master_addr ${MASTER_ADDR:-127.0.0.1} \
+    --master_port ${MASTER_PORT:-26669}
+"
+
+echo $DISTRIBUTED_ARGS
+
+torchrun $DISTRIBUTED_ARGS \
+../../../funasr/bin/train_ds.py \
+++model="${model_name_or_model_dir}" \
+++train_data_set_list="${train_data}" \
+++valid_data_set_list="${val_data}" \
+++dataset="AudioDataset" \
+++dataset_conf.index_ds="IndexDSJsonl" \
+++dataset_conf.data_split_num=1 \
+++dataset_conf.batch_sampler="BatchSampler" \
+++dataset_conf.batch_size=6000  \
+++dataset_conf.sort_size=1024 \
+++dataset_conf.batch_type="token" \
+++dataset_conf.num_workers=4 \
+++train_conf.max_epoch=50 \
+++train_conf.log_interval=1 \
+++train_conf.resume=true \
+++train_conf.validate_interval=2000 \
+++train_conf.save_checkpoint_interval=2000 \
+++train_conf.keep_nbest_models=20 \
+++train_conf.use_deepspeed=false \
+++train_conf.deepspeed_config=${deepspeed_config} \
+++optim_conf.lr=0.0002 \
+++output_dir="${output_dir}" &> ${log_file}
\ No newline at end of file

FunASR/examples/industrial_data_pretraining/campplus_sv/demo.py

+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+from funasr import AutoModel
+
+model = AutoModel(model="iic/speech_campplus_sv_zh-cn_16k-common")
+
+res = model.generate(
+    input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav"
+)
+print(res)

FunASR/examples/industrial_data_pretraining/conformer/demo.py

+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+from funasr import AutoModel
+
+model = AutoModel(model="iic/speech_conformer_asr_nat-zh-cn-16k-aishell2-vocab5212-pytorch")
+
+res = model.generate(
+    input="https://modelscope.oss-cn-beijing.aliyuncs.com/test/audios/asr_example.wav",
+    decoding_ctc_weight=0.0,
+)
+print(res)

FunASR/examples/industrial_data_pretraining/conformer/demo.sh

+
+model="iic/speech_conformer_asr_nat-zh-cn-16k-aishell1-vocab4234-pytorch"
+
+python funasr/bin/inference.py \
++model=${model} \
++input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav" \
++output_dir="./outputs/debug" \
++device="cpu" \
+

FunASR/examples/industrial_data_pretraining/contextual_paraformer/demo.py

+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+from funasr import AutoModel
+
+model = AutoModel(model="iic/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404")
+
+res = model.generate(
+    input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav",
+    hotword="达摩院 魔搭",
+)
+print(res)

FunASR/examples/industrial_data_pretraining/contextual_paraformer/demo.sh

+
+model="iic/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404"
+
+
+python ../../../funasr/bin/inference.py \
++model=${model} \
++input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav" \
++output_dir="./outputs/debug" \
++device="cpu" \
++"hotword='达摩院 魔搭'"
+

FunASR/examples/industrial_data_pretraining/contextual_paraformer/demo2.sh

+python -m funasr.bin.inference \
+--config-path="/nfs/yufan.yf/workspace/model_download/modelscope/hub/damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404" \
+--config-name="config.yaml" \
+++init_param="/nfs/yufan.yf/workspace/model_download/modelscope/hub/damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404/model.pb" \
+++tokenizer_conf.token_list="/nfs/yufan.yf/workspace/model_download/modelscope/hub/damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404/tokens.txt" \
+++frontend_conf.cmvn_file="/nfs/yufan.yf/workspace/model_download/modelscope/hub/damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404/am.mvn" \
+++input="/nfs/yufan.yf/workspace/model_download/modelscope/hub/damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404/asr_example_zh.wav" \
+++output_dir="./outputs/debug2" \
+++device="" \

FunASR/examples/industrial_data_pretraining/contextual_paraformer/finetune.sh

+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+workspace=`pwd`
+
+# method1, finetune from model hub
+
+# which gpu to train or finetune
+export CUDA_VISIBLE_DEVICES="0,1"
+gpu_num=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')
+
+# model_name from model_hub, or model_dir in local path
+
+## option 1, download model automatically
+model_name_or_model_dir="iic/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404"
+
+
+## option 2, download model by git
+#local_path_root=${workspace}/modelscope_models
+#mkdir -p ${local_path_root}/${model_name_or_model_dir}
+#git clone https://www.modelscope.cn/${model_name_or_model_dir}.git ${local_path_root}/${model_name_or_model_dir}
+#model_name_or_model_dir=${local_path_root}/${model_name_or_model_dir}
+
+
+# data dir, which contains: train.json, val.json
+data_dir="../../../data/list"
+
+train_data="${data_dir}/train.jsonl"
+val_data="${data_dir}/val.jsonl"
+
+# generate train.jsonl and val.jsonl from wav.scp and text.txt
+scp2jsonl \
+++scp_file_list='["../../../data/list/train_wav.scp", "../../../data/list/train_text.txt"]' \
+++data_type_list='["source", "target"]' \
+++jsonl_file_out="${train_data}"
+
+scp2jsonl \
+++scp_file_list='["../../../data/list/val_wav.scp", "../../../data/list/val_text.txt"]' \
+++data_type_list='["source", "target"]' \
+++jsonl_file_out="${val_data}"
+
+
+# exp output dir
+output_dir="./outputs"
+log_file="${output_dir}/log.txt"
+
+
+mkdir -p ${output_dir}
+echo "log_file: ${log_file}"
+
+deepspeed_config=${workspace}/../../ds_stage1.json
+
+DISTRIBUTED_ARGS="
+    --nnodes ${WORLD_SIZE:-1} \
+    --nproc_per_node $gpu_num \
+    --node_rank ${RANK:-0} \
+    --master_addr ${MASTER_ADDR:-127.0.0.1} \
+    --master_port ${MASTER_PORT:-26669}
+"
+
+echo $DISTRIBUTED_ARGS
+
+torchrun $DISTRIBUTED_ARGS \
+../../../funasr/bin/train_ds.py \
+++model="${model_name_or_model_dir}" \
+++train_data_set_list="${train_data}" \
+++valid_data_set_list="${val_data}" \
+++dataset="AudioDatasetHotword" \
+++dataset_conf.index_ds="IndexDSJsonl" \
+++dataset_conf.data_split_num=1 \
+++dataset_conf.batch_sampler="BatchSampler" \
+++dataset_conf.batch_size=6000  \
+++dataset_conf.sort_size=1024 \
+++dataset_conf.batch_type="token" \
+++dataset_conf.num_workers=4 \
+++train_conf.max_epoch=50 \
+++train_conf.log_interval=1 \
+++train_conf.resume=true \
+++train_conf.validate_interval=2000 \
+++train_conf.save_checkpoint_interval=2000 \
+++train_conf.keep_nbest_models=20 \
+++train_conf.use_deepspeed=false \
+++train_conf.deepspeed_config=${deepspeed_config} \
+++optim_conf.lr=0.0002 \
+++output_dir="${output_dir}" &> ${log_file}
\ No newline at end of file

FunASR/examples/industrial_data_pretraining/contextual_paraformer/path.sh

+export FUNASR_DIR=$PWD/../../../
+
+# NOTE(kan-bayashi): Use UTF-8 in Python to avoid UnicodeDecodeError when LC_ALL=C
+export PYTHONIOENCODING=UTF-8
+export PATH=$FUNASR_DIR/funasr/bin:$PATH
+export PYTHONPATH=$FUNASR_DIR/funasr/bin:$FUNASR_DIR/funasr:$FUNASR_DIR:$PYTHONPATH

FunASR/examples/industrial_data_pretraining/ct_transformer/demo.py

+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+from funasr import AutoModel
+
+model = AutoModel(model="iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch")
+
+res = model.generate(
+    input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/punc_example.txt"
+)
+print(res)
+
+
+from funasr import AutoModel
+
+model = AutoModel(model="iic/punc_ct-transformer_cn-en-common-vocab471067-large")
+
+res = model.generate(
+    input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/punc_example.txt"
+)
+print(res)

FunASR/examples/industrial_data_pretraining/ct_transformer/demo.sh

+
+#model="iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch"
+#
+
+model="iic/punc_ct-transformer_cn-en-common-vocab471067-large"
+
+
+python funasr/bin/inference.py \
++model=${model} \
++input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/punc_example.txt" \
++output_dir="./outputs/debug" \
++device="cpu"

FunASR/examples/industrial_data_pretraining/ct_transformer/export.py

+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+# method1, inference from model hub
+
+from funasr import AutoModel
+
+model = AutoModel(
+    model="iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch",
+)
+
+res = model.export(type="onnx", quantize=False)
+print(res)
+
+
+# method2, inference from local path
+from funasr import AutoModel
+
+model = AutoModel(
+    model="/Users/zhifu/.cache/modelscope/hub/iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch"
+)
+
+res = model.export(type="onnx", quantize=False)
+print(res)

FunASR/examples/industrial_data_pretraining/ct_transformer/export.sh

+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+# method1, inference from model hub
+export HYDRA_FULL_ERROR=1
+
+
+model="iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch"
+
+
+python -m funasr.bin.export \
+++model=${model} \
+++model_revision=${model_revision} \
+++type="onnx" \
+++quantize=false \
+++device="cpu"
+
+
+# method2, inference from local path
+model="/Users/zhifu/.cache/modelscope/hub/iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch"
+
+python -m funasr.bin.export \
+++model=${model} \
+++type="onnx" \
+++quantize=false \
+++device="cpu"
\ No newline at end of file

FunASR/examples/industrial_data_pretraining/ct_transformer_streaming/demo.py

+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+from funasr import AutoModel
+
+model = AutoModel(model="iic/punc_ct-transformer_zh-cn-common-vad_realtime-vocab272727")
+
+inputs = "跨境河流是养育沿岸|人民的生命之源长期以来为帮助下游地区防灾减灾中方技术人员|在上游地区极为恶劣的自然条件下克服巨大困难甚至冒着生命危险|向印方提供汛期水文资料处理紧急事件中方重视印方在跨境河流问题上的关切|愿意进一步完善双方联合工作机制|凡是|中方能做的我们|都会去做而且会做得更好我请印度朋友们放心中国在上游的|任何开发利用都会经过科学|规划和论证兼顾上下游的利益"
+vads = inputs.split("|")
+rec_result_all = "outputs: "
+cache = {}
+for vad in vads:
+    rec_result = model.generate(input=vad, cache=cache)
+    rec_result_all += rec_result[0]["text"]
+
+print(rec_result_all)

FunASR/examples/industrial_data_pretraining/ct_transformer_streaming/demo.sh

+
+model="iic/punc_ct-transformer_zh-cn-common-vad_realtime-vocab272727"
+
+
+python funasr/bin/inference.py \
++model=${model} \
++input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/punc_example.txt" \
++output_dir="./outputs/debug" \
++device="cpu"

FunASR/examples/industrial_data_pretraining/ct_transformer_streaming/export.py

+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+# method1, inference from model hub
+
+from funasr import AutoModel
+
+model = AutoModel(
+    model="iic/punc_ct-transformer_zh-cn-common-vad_realtime-vocab272727",
+)
+
+res = model.export(type="onnx", quantize=False)
+print(res)
+
+
+# method2, inference from local path
+from funasr import AutoModel
+
+model = AutoModel(
+    model="/Users/zhifu/.cache/modelscope/hub/iic/punc_ct-transformer_zh-cn-common-vad_realtime-vocab272727"
+)
+
+res = model.export(type="onnx", quantize=False)
+print(res)

FunASR/examples/industrial_data_pretraining/ct_transformer_streaming/export.sh

+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+# method1, inference from model hub
+export HYDRA_FULL_ERROR=1
+
+
+model="iic/punc_ct-transformer_zh-cn-common-vad_realtime-vocab272727"
+
+
+python -m funasr.bin.export \
+++model=${model} \
+++model_revision=${model_revision} \
+++input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav" \
+++type="onnx" \
+++quantize=false \
+++device="cpu"
+
+
+# method2, inference from local path
+model="/Users/zhifu/.cache/modelscope/hub/iic/punc_ct-transformer_zh-cn-common-vad_realtime-vocab272727"
+
+python -m funasr.bin.export \
+++model=${model} \
+++input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav" \
+++type="onnx" \
+++quantize=false \
+++device="cpu"
\ No newline at end of file

FunASR/examples/industrial_data_pretraining/ctc/demo.py

+#!/usr/bin/env python3
+# -*- encoding: utf-8 -*-
+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+import sys
+from funasr import AutoModel
+
+
+model_dir = "/Users/zhifu/Downloads/modelscope_models/ctc_model"
+input_file = (
+    "https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav"
+)
+
+model = AutoModel(
+    model=model_dir,
+)
+
+res = model.generate(
+    input=input_file,
+    cache={},
+)
+
+print(res)

FunASR/examples/industrial_data_pretraining/ctc/infer_from_local.sh

+# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
+#  MIT License  (https://opensource.org/licenses/MIT)
+
+# method2, inference from local model
+
+# for more input type, please ref to readme.md
+model_dir=$1
+input_file=$2
+output_dir=$3
+
+# download model
+device="cuda:0" # "cuda:0" for gpu0, "cuda:1" for gpu1, "cpu"
+
+tokens="${model_dir}/tokens.json"
+cmvn_file="${model_dir}/am.mvn"
+
+config="config.yaml"
+init_param="${model_dir}/model.pt"
+
+mkdir -p ${output_dir}
+
+python -m funasr.bin.inference \
+--config-path "${model_dir}" \
+--config-name "${config}" \
+++init_param="${init_param}" \
+++tokenizer_conf.token_list="${tokens}" \
+++frontend_conf.cmvn_file="${cmvn_file}" \
+++input="${input_file}" \
+++output_dir="${output_dir}" \
+++device="${device}" \
+