Add non-streaming ASR (#92)

.github/scripts/Main.kt

@@ -33,18 +33,20 @@ fun main() {
         config = config,
     )
 
-    var samples = WaveReader.readWave(
+    var objArray = WaveReader.readWave(
         assetManager = AssetManager(),
         filename = "./sherpa-onnx-streaming-zipformer-en-2023-02-21/test_wavs/1089-134686-0001.wav",
     )
+    var samples : FloatArray = objArray[0] as FloatArray
+    var sampleRate : Int = objArray[1] as Int
 
-    model.acceptWaveform(samples!!, sampleRate=16000)
+    model.acceptWaveform(samples, sampleRate=sampleRate)
     while (model.isReady()) {
       model.decode()
     }
 
-    var tail_paddings = FloatArray(8000) // 0.5 seconds
-    model.acceptWaveform(tail_paddings, sampleRate=16000)
+    var tail_paddings = FloatArray((sampleRate * 0.5).toInt()) // 0.5 seconds
+    model.acceptWaveform(tail_paddings, sampleRate=sampleRate)
     model.inputFinished()
     while (model.isReady()) {
       model.decode()

.github/scripts/test-offline-transducer.sh

@@ -0,0 +1,60 @@
+#!/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]}) $*"
+}
+
+echo "EXE is $EXE"
+echo "PATH: $PATH"
+
+which $EXE
+
+log "------------------------------------------------------------"
+log "Run Conformer transducer (English)"
+log "------------------------------------------------------------"
+
+repo_url=https://huggingface.co/csukuangfj/sherpa-onnx-conformer-en-2023-03-18
+log "Start testing ${repo_url}"
+repo=$(basename $repo_url)
+log "Download pretrained model and test-data from $repo_url"
+
+GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
+pushd $repo
+git lfs pull --include "*.onnx"
+cd test_wavs
+popd
+
+waves=(
+$repo/test_wavs/0.wav
+$repo/test_wavs/1.wav
+$repo/test_wavs/2.wav
+)
+
+for wave in ${waves[@]}; do
+  time $EXE \
+  $repo/tokens.txt \
+  $repo/encoder-epoch-99-avg-1.onnx \
+  $repo/decoder-epoch-99-avg-1.onnx \
+  $repo/joiner-epoch-99-avg-1.onnx \
+  $wave \
+  2
+done
+
+
+if command -v sox &> /dev/null; then
+  echo "test 8kHz"
+  sox $repo/test_wavs/0.wav -r 8000 8k.wav
+  time $EXE \
+    $repo/tokens.txt \
+    $repo/encoder-epoch-99-avg-1.onnx \
+    $repo/decoder-epoch-99-avg-1.onnx \
+    $repo/joiner-epoch-99-avg-1.onnx \
+    8k.wav \
+    2
+fi
+
+rm -rf $repo

.github/scripts/test-online-transducer.sh

@@ -40,7 +40,7 @@ for wave in ${waves[@]}; do
   $repo/decoder-epoch-99-avg-1.onnx \
   $repo/joiner-epoch-99-avg-1.onnx \
   $wave \
-  4
+  2
 done
 
 rm -rf $repo
@@ -72,7 +72,7 @@ for wave in ${waves[@]}; do
   $repo/decoder-epoch-11-avg-1.onnx \
   $repo/joiner-epoch-11-avg-1.onnx \
   $wave \
-  4
+  2
 done
 
 rm -rf $repo
@@ -104,7 +104,7 @@ for wave in ${waves[@]}; do
   $repo/decoder-epoch-99-avg-1.onnx \
   $repo/joiner-epoch-99-avg-1.onnx \
   $wave \
-  4
+  2
 done
 
 rm -rf $repo
@@ -138,7 +138,7 @@ for wave in ${waves[@]}; do
   $repo/decoder-epoch-99-avg-1.onnx \
   $repo/joiner-epoch-99-avg-1.onnx \
   $wave \
-  4
+  2
 done
 
 # Decode a URL
@@ -149,7 +149,7 @@ if [ $EXE == "sherpa-onnx-ffmpeg" ]; then
   $repo/decoder-epoch-99-avg-1.onnx \
   $repo/joiner-epoch-99-avg-1.onnx \
   https://huggingface.co/csukuangfj/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/resolve/main/test_wavs/4.wav \
-  4
+  2
 fi
 
 rm -rf $repo

.github/workflows/linux.yaml

@@ -7,11 +7,11 @@ on:
     paths:
       - '.github/workflows/linux.yaml'
       - '.github/scripts/test-online-transducer.sh'
+      - '.github/scripts/test-offline-transducer.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
       - 'sherpa-onnx/c-api/*'
-      - 'ffmpeg-examples/**'
       - 'c-api-examples/**'
   pull_request:
     branches:
@@ -19,11 +19,11 @@ on:
     paths:
       - '.github/workflows/linux.yaml'
       - '.github/scripts/test-online-transducer.sh'
+      - '.github/scripts/test-offline-transducer.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
       - 'sherpa-onnx/c-api/*'
-      - 'ffmpeg-examples/**'
 
 concurrency:
   group: linux-${{ github.ref }}
@@ -39,35 +39,26 @@ jobs:
       fail-fast: false
       matrix:
         os: [ubuntu-latest]
+        build_type: [Release, Debug]
 
     steps:
       - uses: actions/checkout@v2
         with:
           fetch-depth: 0
 
-      - name: Install ffmpeg
+      - name: Install sox
         shell: bash
         run: |
-          sudo apt-get install -y software-properties-common
-          sudo add-apt-repository ppa:savoury1/ffmpeg4
-          sudo add-apt-repository ppa:savoury1/ffmpeg5
-
-          sudo apt-get install -y libavdevice-dev libavutil-dev ffmpeg
-          pkg-config --modversion libavutil
-          ffmpeg -version
-
-      - name: Show ffmpeg version
-        shell: bash
-        run: |
-          pkg-config --modversion libavutil
-          ffmpeg -version
+          sudo apt-get update
+          sudo apt-get install -y sox
+          sox -h
 
       - name: Configure CMake
         shell: bash
         run: |
           mkdir build
           cd build
-          cmake -D CMAKE_BUILD_TYPE=Release ..
+          cmake -D CMAKE_BUILD_TYPE=${{ matrix.build_type }} ..
 
       - name: Build sherpa-onnx for ubuntu
         shell: bash
@@ -78,21 +69,19 @@ jobs:
           ls -lh lib
           ls -lh bin
 
-          cd ../ffmpeg-examples
-          make
-
       - name: Display dependencies of sherpa-onnx for linux
         shell: bash
         run: |
           file build/bin/sherpa-onnx
           readelf -d build/bin/sherpa-onnx
 
-      - name: Test sherpa-onnx-ffmpeg
+      - name: Test offline transducer
+        shell: bash
         run: |
-          export PATH=$PWD/ffmpeg-examples:$PATH
-          export EXE=sherpa-onnx-ffmpeg
+          export PATH=$PWD/build/bin:$PATH
+          export EXE=sherpa-onnx-offline
 
-          .github/scripts/test-online-transducer.sh
+          .github/scripts/test-offline-transducer.sh
 
       - name: Test online transducer
         shell: bash

.github/workflows/macos.yaml

@@ -7,6 +7,7 @@ on:
     paths:
       - '.github/workflows/macos.yaml'
       - '.github/scripts/test-online-transducer.sh'
+      - '.github/scripts/test-offline-transducer.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -16,6 +17,7 @@ on:
     paths:
       - '.github/workflows/macos.yaml'
       - '.github/scripts/test-online-transducer.sh'
+      - '.github/scripts/test-offline-transducer.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -34,18 +36,25 @@ jobs:
       fail-fast: false
       matrix:
         os: [macos-latest]
+        build_type: [Release, Debug]
 
     steps:
       - uses: actions/checkout@v2
         with:
           fetch-depth: 0
 
+      - name: Install sox
+        shell: bash
+        run: |
+          brew install sox
+          sox -h
+
       - name: Configure CMake
         shell: bash
         run: |
           mkdir build
           cd build
-          cmake -D CMAKE_BUILD_TYPE=Release ..
+          cmake -D CMAKE_BUILD_TYPE=${{ matrix.build_type }} ..
 
       - name: Build sherpa-onnx for macos
         shell: bash
@@ -64,6 +73,14 @@ jobs:
           otool -L build/bin/sherpa-onnx
           otool -l build/bin/sherpa-onnx
 
+      - name: Test offline transducer
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin:$PATH
+          export EXE=sherpa-onnx-offline
+
+          .github/scripts/test-offline-transducer.sh
+
       - name: Test online transducer
         shell: bash
         run: |