Add Android demo for spoken language identification using Whisper multilingual models (#783)

scripts/apk/build-apk-slid.sh.in

@@ -0,0 +1,91 @@
+#!/usr/bin/env bash
+#
+# Auto generated! Please DO NOT EDIT!
+
+# Please set the environment variable ANDROID_NDK
+# before running this script
+
+# Inside the $ANDROID_NDK directory, you can find a binary ndk-build
+# and some other files like the file "build/cmake/android.toolchain.cmake"
+
+set -ex
+
+log() {
+  # This function is from espnet
+  local fname=${BASH_SOURCE[1]##*/}
+  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
+}
+
+SHERPA_ONNX_VERSION=$(grep "SHERPA_ONNX_VERSION" ./CMakeLists.txt  | cut -d " " -f 2  | cut -d '"' -f 2)
+
+log "Building spoken language identification APK for sherpa-onnx v${SHERPA_ONNX_VERSION}"
+
+log "====================arm64-v8a================="
+./build-android-arm64-v8a.sh
+log "====================armv7-eabi================"
+./build-android-armv7-eabi.sh
+log "====================x86-64===================="
+./build-android-x86-64.sh
+log "====================x86===================="
+./build-android-x86.sh
+
+mkdir -p apks
+
+{% for model in model_list %}
+pushd ./android/SherpaOnnxSpokenLanguageIdentification/app/src/main/assets/
+model_name={{ model.model_name }}
+short_name={{ model.short_name }}
+type={{ model.idx }}
+
+curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/${model_name}.tar.bz2
+tar xvf ${model_name}.tar.bz2
+rm -rfv $model_name/*-encoder.onnx
+rm -rfv $model_name/*-decoder.onnx
+rm -rfv $model_name/*.py
+rm -rfv $model_name/*.txt
+rm -rfv $model_name/*.md
+rm -rfv $model_name/test_wavs
+rm -rf  *.tar.bz2
+ls -lh $model_name
+
+popd
+# Now we are at the project root directory
+
+git checkout .
+pushd android/SherpaOnnxSpokenLanguageIdentification/app/src/main/java/com/k2fsa/sherpa/onnx/slid/
+sed -i.bak s/"type = 0/type = $type/" ./slid.kt
+git diff
+popd
+
+for arch in arm64-v8a armeabi-v7a x86_64 x86; do
+  log "------------------------------------------------------------"
+  log "build spoken language identification apk for $arch"
+  log "------------------------------------------------------------"
+  src_arch=$arch
+  if [ $arch == "armeabi-v7a" ]; then
+    src_arch=armv7-eabi
+  elif [ $arch == "x86_64" ]; then
+    src_arch=x86-64
+  fi
+
+  ls -lh ./build-android-$src_arch/install/lib/*.so
+
+  cp -v ./build-android-$src_arch/install/lib/*.so ./android/SherpaOnnxSpokenLanguageIdentification/app/src/main/jniLibs/$arch/
+
+  pushd ./android/SherpaOnnxSpokenLanguageIdentification
+  sed -i.bak s/2048/9012/g ./gradle.properties
+  git diff ./gradle.properties
+  ./gradlew assembleRelease
+  popd
+
+  mv android/SherpaOnnxSpokenLanguageIdentification/app/build/outputs/apk/release/app-release-unsigned.apk ./apks/sherpa-onnx-${SHERPA_ONNX_VERSION}-$arch-audio-tagging-$short_name.apk
+  ls -lh apks
+  rm -v ./android/SherpaOnnxSpokenLanguageIdentification/app/src/main/jniLibs/$arch/*.so
+done
+
+rm -rf ./android/SherpaOnnxSpokenLanguageIdentification/app/src/main/assets/$model_name
+{% endfor %}
+
+git checkout .
+
+ls -lh apks/

scripts/apk/generate-slid-apk-script.py

@@ -0,0 +1,90 @@
+#!/usr/bin/env python3
+
+import argparse
+from dataclasses import dataclass
+from typing import List, Optional
+
+import jinja2
+
+
+def get_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--total",
+        type=int,
+        default=1,
+        help="Number of runners",
+    )
+    parser.add_argument(
+        "--index",
+        type=int,
+        default=0,
+        help="Index of the current runner",
+    )
+    return parser.parse_args()
+
+
+@dataclass
+class SlidModel:
+    model_name: str
+    idx: int
+    short_name: str = ""
+
+
+def get_models():
+    # see https://k2-fsa.github.io/sherpa/onnx/spolken-language-identification/pretrained_models.html#pre-trained-models
+    whisper_models = [
+        SlidModel(
+            model_name="sherpa-onnx-whisper-tiny",
+            idx=0,
+            short_name="whisper_tiny",
+        ),
+    ]
+
+    return whisper_models
+
+
+def main():
+    args = get_args()
+    index = args.index
+    total = args.total
+    assert 0 <= index < total, (index, total)
+
+    all_model_list = get_models()
+
+    num_models = len(all_model_list)
+
+    num_per_runner = num_models // total
+    if num_per_runner <= 0:
+        raise ValueError(f"num_models: {num_models}, num_runners: {total}")
+
+    start = index * num_per_runner
+    end = start + num_per_runner
+
+    remaining = num_models - args.total * num_per_runner
+
+    print(f"{index}/{total}: {start}-{end}/{num_models}")
+
+    d = dict()
+    d["model_list"] = all_model_list[start:end]
+    if index < remaining:
+        s = args.total * num_per_runner + index
+        d["model_list"].append(all_model_list[s])
+        print(f"{s}/{num_models}")
+
+    filename_list = [
+        "./build-apk-slid.sh",
+    ]
+    for filename in filename_list:
+        environment = jinja2.Environment()
+        with open(f"{filename}.in") as f:
+            s = f.read()
+        template = environment.from_string(s)
+
+        s = template.render(**d)
+        with open(filename, "w") as f:
+            print(s, file=f)
+
+
+if __name__ == "__main__":
+    main()