Add Android demo for speaker recognition (#536)

See pre-built Android APKs at 
https://k2-fsa.github.io/sherpa/onnx/speaker-identification/apk.html

kotlin-api-examples/Main.kt

@@ -7,11 +7,67 @@ fun callback(samples: FloatArray): Unit {
 }
 
 fun main() {
+  testSpeakerRecognition()
   testTts()
   testAsr("transducer")
   testAsr("zipformer2-ctc")
 }
 
+fun computeEmbedding(extractor: SpeakerEmbeddingExtractor, filename: String): FloatArray {
+    var objArray = WaveReader.readWaveFromFile(
+        filename = filename,
+    )
+    var samples: FloatArray = objArray[0] as FloatArray
+    var sampleRate: Int = objArray[1] as Int
+
+    val stream = extractor.createStream()
+    stream.acceptWaveform(sampleRate = sampleRate, samples=samples)
+    stream.inputFinished()
+    check(extractor.isReady(stream))
+
+    val embedding = extractor.compute(stream)
+
+    stream.release()
+
+    return embedding
+}
+
+fun testSpeakerRecognition() {
+    val config = SpeakerEmbeddingExtractorConfig(
+        model="./3dspeaker_speech_eres2net_large_sv_zh-cn_3dspeaker_16k.onnx",
+        )
+    val extractor = SpeakerEmbeddingExtractor(config = config)
+
+    val embedding1a = computeEmbedding(extractor, "./speaker1_a_cn_16k.wav")
+    val embedding2a = computeEmbedding(extractor, "./speaker2_a_cn_16k.wav")
+    val embedding1b = computeEmbedding(extractor, "./speaker1_b_cn_16k.wav")
+
+    var manager = SpeakerEmbeddingManager(extractor.dim())
+    var ok = manager.add(name = "speaker1", embedding=embedding1a)
+    check(ok)
+
+    manager.add(name = "speaker2", embedding=embedding2a)
+    check(ok)
+
+    var name = manager.search(embedding=embedding1b, threshold=0.5f)
+    check(name == "speaker1")
+
+    manager.release()
+
+    manager = SpeakerEmbeddingManager(extractor.dim())
+    val embeddingList = mutableListOf(embedding1a, embedding1b)
+    ok = manager.add(name = "s1", embedding=embeddingList.toTypedArray())
+    check(ok)
+
+    name = manager.search(embedding=embedding1b, threshold=0.5f)
+    check(name == "s1")
+
+    name = manager.search(embedding=embedding2a, threshold=0.5f)
+    check(name.length == 0)
+
+    manager.release()
+}
+
 fun testTts() {
   // see https://github.com/k2-fsa/sherpa-onnx/releases/tag/tts-models
   // https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-amy-low.tar.bz2

kotlin-api-examples/Speaker.kt

@@ -0,0 +1 @@
+../android/SherpaOnnxSpeakerIdentification/app/src/main/java/com/k2fsa/sherpa/onnx/speaker/identification/Speaker.kt

kotlin-api-examples/run.sh

@@ -29,6 +29,22 @@ export LD_LIBRARY_PATH=$PWD/build/lib:$LD_LIBRARY_PATH
 
 cd ../kotlin-api-examples
 
+if [ ! -f ./3dspeaker_speech_eres2net_large_sv_zh-cn_3dspeaker_16k.onnx ]; then
+  wget -q https://github.com/k2-fsa/sherpa-onnx/releases/download/speaker-recongition-models/3dspeaker_speech_eres2net_large_sv_zh-cn_3dspeaker_16k.onnx
+fi
+
+if [ ! -f ./speaker1_a_cn_16k.wav ]; then
+  wget -q https://github.com/csukuangfj/sr-data/raw/main/test/3d-speaker/speaker1_a_cn_16k.wav
+fi
+
+if [ ! -f ./speaker1_b_cn_16k.wav ]; then
+  wget -q https://github.com/csukuangfj/sr-data/raw/main/test/3d-speaker/speaker1_b_cn_16k.wav
+fi
+
+if [ ! -f ./speaker2_a_cn_16k.wav ]; then
+  wget -q https://github.com/csukuangfj/sr-data/raw/main/test/3d-speaker/speaker2_a_cn_16k.wav
+fi
+
 if [ ! -f ./sherpa-onnx-streaming-zipformer-en-2023-02-21/tokens.txt ]; then
   git lfs install
   git clone https://huggingface.co/csukuangfj/sherpa-onnx-streaming-zipformer-en-2023-02-21
@@ -46,7 +62,7 @@ if [ ! -f ./vits-piper-en_US-amy-low/en_US-amy-low.onnx ]; then
   rm vits-piper-en_US-amy-low.tar.bz2
 fi
 
-kotlinc-jvm -include-runtime -d main.jar Main.kt WaveReader.kt SherpaOnnx.kt faked-asset-manager.kt Tts.kt
+kotlinc-jvm -include-runtime -d main.jar Main.kt WaveReader.kt SherpaOnnx.kt faked-asset-manager.kt Tts.kt Speaker.kt
 
 ls -lh main.jar