Migrate zipformer model to other Chinese datasets (#1216)

added zipformer recipe for AISHELL-1

.github/scripts/run-aishell-pruned-transducer-stateless3-2022-06-20.sh

@@ -18,8 +18,8 @@ log "Downloading pre-commputed fbank from $fbank_url"
 git clone https://huggingface.co/csukuangfj/aishell-test-dev-manifests
 ln -s $PWD/aishell-test-dev-manifests/data .
 
-log "Downloading pre-trained model from $repo_url"
 repo_url=https://huggingface.co/csukuangfj/icefall-aishell-pruned-transducer-stateless3-2022-06-20
+log "Downloading pre-trained model from $repo_url"
 git clone $repo_url
 repo=$(basename $repo_url)
 

.github/scripts/run-aishell-zipformer-2023-10-24.sh

@@ -0,0 +1,103 @@
+#!/usr/bin/env bash
+
+set -e
+
+log() {
+  # This function is from espnet
+  local fname=${BASH_SOURCE[1]##*/}
+  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
+}
+
+cd egs/aishell/ASR
+
+git lfs install
+
+fbank_url=https://huggingface.co/csukuangfj/aishell-test-dev-manifests
+log "Downloading pre-commputed fbank from $fbank_url"
+
+git clone https://huggingface.co/csukuangfj/aishell-test-dev-manifests
+ln -s $PWD/aishell-test-dev-manifests/data .
+
+log "======================="
+log "CI testing large model"
+repo_url=https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-large-2023-10-24/
+log "Downloading pre-trained model from $repo_url"
+git clone $repo_url
+repo=$(basename $repo_url)
+
+log "Display test files"
+tree $repo/
+ls -lh $repo/test_wavs/*.wav
+
+for method in modified_beam_search greedy_search fast_beam_search; do
+  log "$method"
+
+  ./zipformer/pretrained.py \
+    --method $method \
+    --context-size 1 \
+    --checkpoint $repo/exp/pretrained.pt \
+    --tokens $repo/data/lang_char/tokens.txt \
+    --num-encoder-layers 2,2,4,5,4,2 \
+    --feedforward-dim 512,768,1536,2048,1536,768 \
+    --encoder-dim 192,256,512,768,512,256 \
+    --encoder-unmasked-dim 192,192,256,320,256,192 \
+    $repo/test_wavs/BAC009S0764W0121.wav \
+    $repo/test_wavs/BAC009S0764W0122.wav \
+    $repo/test_wavs/BAC009S0764W0123.wav
+done
+
+log "======================="
+log "CI testing medium model"
+repo_url=https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-2023-10-24/
+log "Downloading pre-trained model from $repo_url"
+git clone $repo_url
+repo=$(basename $repo_url)
+
+log "Display test files"
+tree $repo/
+ls -lh $repo/test_wavs/*.wav
+
+
+for method in modified_beam_search greedy_search fast_beam_search; do
+  log "$method"
+
+  ./zipformer/pretrained.py \
+    --method $method \
+    --context-size 1 \
+    --checkpoint $repo/exp/pretrained.pt \
+    --tokens $repo/data/lang_char/tokens.txt \
+    $repo/test_wavs/BAC009S0764W0121.wav \
+    $repo/test_wavs/BAC009S0764W0122.wav \
+    $repo/test_wavs/BAC009S0764W0123.wav
+done
+
+
+log "======================="
+log "CI testing small model"
+repo_url=https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-small-2023-10-24/
+log "Downloading pre-trained model from $repo_url"
+git clone $repo_url
+repo=$(basename $repo_url)
+
+log "Display test files"
+tree $repo/
+ls -lh $repo/test_wavs/*.wav
+
+
+for method in modified_beam_search greedy_search fast_beam_search; do
+  log "$method"
+
+  ./zipformer/pretrained.py \
+    --method $method \
+    --context-size 1 \
+    --checkpoint $repo/exp/pretrained.pt \
+    --tokens $repo/data/lang_char/tokens.txt \
+    --num-encoder-layers 2,2,2,2,2,2 \
+    --feedforward-dim 512,768,768,768,768,768 \
+    --encoder-dim 192,256,256,256,256,256 \
+    --encoder-unmasked-dim 192,192,192,192,192,192 \
+    $repo/test_wavs/BAC009S0764W0121.wav \
+    $repo/test_wavs/BAC009S0764W0122.wav \
+    $repo/test_wavs/BAC009S0764W0123.wav
+done
+

.github/workflows/run-aishell-zipformer-2023-10-24.yml

@@ -0,0 +1,95 @@
+# Copyright      2023  Zengrui Jin (Xiaomi Corp.)
+
+# See ../../LICENSE for clarification regarding multiple authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: run-aishell-zipformer-2023-10-24
+
+on:
+  push:
+    branches:
+      - master
+  pull_request:
+    types: [labeled]
+
+  schedule:
+    # minute (0-59)
+    # hour (0-23)
+    # day of the month (1-31)
+    # month (1-12)
+    # day of the week (0-6)
+    # nightly build at 15:50 UTC time every day
+    - cron: "50 15 * * *"
+
+concurrency:
+  group: run_aishell_zipformer_2023_10_24-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  run_aishell_zipformer_2023_10_24:
+    if: github.event.label.name == 'ready' || github.event.label.name == 'zipformer' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ubuntu-latest]
+        python-version: [3.8]
+
+      fail-fast: false
+
+    steps:
+      - uses: actions/checkout@v2
+        with:
+          fetch-depth: 0
+
+      - name: Setup Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v2
+        with:
+          python-version: ${{ matrix.python-version }}
+          cache: 'pip'
+          cache-dependency-path: '**/requirements-ci.txt'
+
+      - name: Install Python dependencies
+        run: |
+          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
+          pip uninstall -y protobuf
+          pip install --no-binary protobuf protobuf==3.20.*
+
+      - name: Cache kaldifeat
+        id: my-cache
+        uses: actions/cache@v2
+        with:
+          path: |
+            ~/tmp/kaldifeat
+          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
+
+      - name: Install kaldifeat
+        if: steps.my-cache.outputs.cache-hit != 'true'
+        shell: bash
+        run: |
+          .github/scripts/install-kaldifeat.sh
+
+      - name: Inference with pre-trained model
+        shell: bash
+        env:
+          GITHUB_EVENT_NAME: ${{ github.event_name }}
+          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
+        run: |
+          sudo apt-get -qq install git-lfs tree
+          export PYTHONPATH=$PWD:$PYTHONPATH
+          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
+          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
+
+          .github/scripts/run-aishell-zipformer-2023-10-24.sh
+
+      

egs/aidatatang_200zh/ASR/prepare.sh

@@ -7,6 +7,8 @@ set -eou pipefail
 
 stage=-1
 stop_stage=100
+perturb_speed=true
+
 
 # We assume dl_dir (download dir) contains the following
 # directories and files. If not, they will be downloaded
@@ -77,7 +79,7 @@ if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
   log "Stage 4: Compute fbank for aidatatang_200zh"
   if [ ! -f data/fbank/.aidatatang_200zh.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_aidatatang_200zh.py --perturb-speed True
+    ./local/compute_fbank_aidatatang_200zh.py --perturb-speed ${perturb_speed}
     touch data/fbank/.aidatatang_200zh.done
   fi
 fi

egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/asr_datamodule.py

@@ -102,7 +102,7 @@ def add_arguments(cls, parser: argparse.ArgumentParser):
         group.add_argument(
             "--bucketing-sampler",
             type=str2bool,
-            default=True,
+            default=False,
             help="When enabled, the batches will come from buckets of "
             "similar duration (saves padding frames).",
         )
@@ -289,6 +289,7 @@ def train_dataloaders(
                 shuffle=self.args.shuffle,
                 num_buckets=self.args.num_buckets,
                 drop_last=True,
+                buffer_size=50000,
             )
         else:
             logging.info("Using SimpleCutSampler.")

egs/aishell/ASR/README.md

@@ -1,10 +1,12 @@
 
 # Introduction
 
-Please refer to <https://icefall.readthedocs.io/en/latest/recipes/Non-streaming-ASR/aishell/index.html>
-for how to run models in this recipe.
+Please refer to <https://k2-fsa.github.io/icefall/recipes/Non-streaming-ASR/aishell/index.html> for how to run models in this recipe.
 
+Aishell is an open-source Chinese Mandarin speech corpus published by Beijing Shell Shell Technology Co., Ltd.
+400 people from different accent areas in China are invited to participate in the recording, which is conducted in a quiet indoor environment using high fidelity microphone and downsampled to 16kHz. The manual transcription accuracy is above 95%, through professional speech annotation and strict quality inspection. The data is free for academic use. We hope to provide moderate amount of data for new researchers in the field of speech recognition.
 
+(From [Open Speech and Language Resources](https://www.openslr.org/33/))
 
 # Transducers