Merge pull request #776 from alibaba-damo-academy/dev_cmz

Dev cmz

egs_modelscope/asr_vad_punc/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch/demo.py

@@ -8,7 +8,7 @@ if __name__ == '__main__':
         task=Tasks.auto_speech_recognition,
         model='damo/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch',
         vad_model='damo/speech_fsmn_vad_zh-cn-16k-common-pytorch',
-        punc_model='damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch',
+        punc_model='damo/punc_ct-transformer_cn-en-common-vocab471067-large',
         output_dir=output_dir,
     )
     rec_result = inference_pipeline(audio_in=audio_in, batch_size_token=5000)

egs_modelscope/punctuation/TEMPLATE/infer.py

@@ -15,9 +15,9 @@ def modelscope_infer(args):
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument('--model', type=str, default="damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch")
+    parser.add_argument('--model', type=str, default="damo/punc_ct-transformer_cn-en-common-vocab471067-large")
     parser.add_argument('--text_in', type=str, default="./data/test/punc.txt")
     parser.add_argument('--output_dir', type=str, default="./results/")
     parser.add_argument('--gpuid', type=str, default="0")
     args = parser.parse_args()
-    modelscope_infer(args)
+    modelscope_infer(args)

egs_modelscope/punctuation/TEMPLATE/infer.sh

@@ -7,7 +7,7 @@ set -o pipefail
 stage=1
 stop_stage=2
 model="damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch"
-data_dir="./data/test"
+data_dir="./data"
 output_dir="./results"
 gpu_inference=true    # whether to perform gpu decoding
 gpuid_list="0,1"    # set gpus, e.g., gpuid_list="0,1"
@@ -32,7 +32,7 @@ split_scps=""
 for JOB in $(seq ${nj}); do
     split_scps="$split_scps $output_dir/split/text.$JOB.scp"
 done
-perl utils/split_scp.pl ${data_dir}/punc.txt ${split_scps}
+perl utils/split_scp.pl ${data_dir}/punc_example.txt ${split_scps}
 
 if [ -n "${checkpoint_dir}" ]; then
   python utils/prepare_checkpoint.py ${model} ${checkpoint_dir} ${checkpoint_name}

egs_modelscope/punctuation/punc_ct-transformer_cn-en-common-vocab471067-large/README.md

@@ -0,0 +1 @@
+../TEMPLATE/README.md

egs_modelscope/punctuation/punc_ct-transformer_cn-en-common-vocab471067-large/data/punc_example.txt

@@ -0,0 +1,3 @@
+1	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
+2	从存储上来说仅仅是全景图片它就会是图片的四倍的容量然后全景的视频会是普通视频八倍的这个存储的容要求而三d的模型会是图片的十倍这都对我们今天运行在的云计算的平台存储的平台提出了更高的要求
+3	那今天的会就到这里吧 happy new year 明年见

egs_modelscope/punctuation/punc_ct-transformer_cn-en-common-vocab471067-large/demo.py

@@ -0,0 +1,22 @@
+
+from modelscope.pipelines import pipeline
+from modelscope.utils.constant import Tasks
+
+inference_pipeline = pipeline(
+    task=Tasks.punctuation,
+    model='damo/punc_ct-transformer_cn-en-common-vocab471067-large',
+    model_revision="v1.0.0",
+    output_dir="./tmp/"
+)
+
+##################text.scp###################
+# inputs = "./egs_modelscope/punctuation/punc_ct-transformer_cn-en-common-vocab471067-large/data/punc_example.txt"
+
+##################text#####################
+#inputs = "我们都是木头人不会讲话不会动"
+
+##################text file url#######################
+inputs = "https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/punc_example.txt"
+
+rec_result = inference_pipeline(text_in=inputs)
+print(rec_result)

egs_modelscope/punctuation/punc_ct-transformer_cn-en-common-vocab471067-large/infer.py

@@ -0,0 +1,25 @@
+import os
+import shutil
+import argparse
+from modelscope.pipelines import pipeline
+from modelscope.utils.constant import Tasks
+
+def modelscope_infer(args):
+    os.environ['CUDA_VISIBLE_DEVICES'] = str(args.gpuid)
+    inference_pipeline = pipeline(
+        task=Tasks.punctuation,
+        model=args.model,
+        model_revision=args.model_revision,        
+        output_dir=args.output_dir,
+    )
+    inference_pipeline(text_in=args.text_in)
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--model', type=str, default="damo/punc_ct-transformer_cn-en-common-vocab471067-large")
+    parser.add_argument('--text_in', type=str, default="./data/test/punc.txt")
+    parser.add_argument('--model_revision', type=str, default=None)
+    parser.add_argument('--output_dir', type=str, default="./results/")
+    parser.add_argument('--gpuid', type=str, default="0")
+    args = parser.parse_args()
+    modelscope_infer(args)

egs_modelscope/punctuation/punc_ct-transformer_cn-en-common-vocab471067-large/infer.sh

@@ -0,0 +1,68 @@
+#!/usr/bin/env bash
+
+set -e
+set -u
+set -o pipefail
+
+stage=1
+stop_stage=2
+model="damo/punc_ct-transformer_cn-en-common-vocab471067-large"
+model_revision="v1.0.0"
+data_dir="./data"
+output_dir="./results"
+gpu_inference=true    # whether to perform gpu decoding
+gpuid_list="0,1"    # set gpus, e.g., gpuid_list="0,1"
+njob=64    # the number of jobs for CPU decoding, if gpu_inference=false, use CPU decoding, please set njob
+checkpoint_dir=
+checkpoint_name="punc.pb"
+
+. utils/parse_options.sh || exit 1;
+
+if ${gpu_inference} == "true"; then
+    nj=$(echo $gpuid_list | awk -F "," '{print NF}')
+else
+    nj=$njob
+    gpuid_list=""
+    for JOB in $(seq ${nj}); do
+        gpuid_list=$gpuid_list"-1,"
+    done
+fi
+
+mkdir -p $output_dir/split
+split_scps=""
+for JOB in $(seq ${nj}); do
+    split_scps="$split_scps $output_dir/split/text.$JOB.scp"
+done
+perl utils/split_scp.pl ${data_dir}/punc_example.txt ${split_scps}
+
+if [ -n "${checkpoint_dir}" ]; then
+  python utils/prepare_checkpoint.py ${model} ${checkpoint_dir} ${checkpoint_name}
+  model=${checkpoint_dir}/${model}
+fi
+
+if [ $stage -le 1 ] && [ $stop_stage -ge 1 ];then
+    echo "Decoding ..."
+    gpuid_list_array=(${gpuid_list//,/ })
+    for JOB in $(seq ${nj}); do
+        {
+        id=$((JOB-1))
+        gpuid=${gpuid_list_array[$id]}
+        mkdir -p ${output_dir}/output.$JOB
+        python infer.py \
+            --model ${model} \
+            --text_in ${output_dir}/split/text.$JOB.scp \
+            --output_dir ${output_dir}/output.$JOB \
+            --model_revision ${model_revision}
+            --gpuid ${gpuid}
+        }&
+    done
+    wait
+
+    mkdir -p ${output_dir}/final_res
+    if [ -f "${output_dir}/output.1/infer.out" ]; then
+      for i in $(seq "${nj}"); do
+          cat "${output_dir}/output.${i}/infer.out"
+      done | sort -k1 >"${output_dir}/final_res/infer.out"
+    fi
+fi
+

egs_modelscope/punctuation/punc_ct-transformer_cn-en-common-vocab471067-large/utils

@@ -0,0 +1 @@
+../../../egs/aishell/transformer/utils

egs_modelscope/punctuation/punc_ct-transformer_zh-cn-common-vocab272727-pytorch/data/punc_example.txt

@@ -0,0 +1,3 @@
+1	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
+2	从存储上来说仅仅是全景图片它就会是图片的四倍的容量然后全景的视频会是普通视频八倍的这个存储的容要求而三d的模型会是图片的十倍这都对我们今天运行在的云计算的平台存储的平台提出了更高的要求
+3	那今天的会就到这里吧 happy new year 明年见

egs_modelscope/punctuation/punc_ct-transformer_zh-cn-common-vocab272727-pytorch/utils

@@ -0,0 +1 @@
+../../../egs/aishell/transformer/utils

funasr/datasets/preprocessor.py

@@ -11,7 +11,7 @@ from typing import Union
 import numpy as np
 import scipy.signal
 import soundfile
-
+import jieba
 
 from funasr.text.build_tokenizer import build_tokenizer
 from funasr.text.cleaner import TextCleaner
@@ -659,7 +659,6 @@ class CodeMixTokenizerCommonPreprocessor(CommonPreprocessor):
         self.split_text_name = split_text_name
         self.seg_jieba = seg_jieba
         if self.seg_jieba:
-            import jieba
             jieba.load_userdict(seg_dict_file)
 
     @classmethod