Add go-api-examples (#219)

go-api-examples/real-time-speech-recognition-from-microphone/.gitignore

@@ -0,0 +1 @@
+real-time-speech-recognition-from-microphone

go-api-examples/real-time-speech-recognition-from-microphone/README.md

@@ -0,0 +1,23 @@
+# Introduction
+
+This examples shows how to use the golang package of [sherpa-onnx][sherpa-onnx]
+for real-time speech recognition from microphone.
+
+It uses <https://github.com/gordonklaus/portaudio>
+to read the microphone and you have to install `portaudio` first.
+
+On macOS, you can use
+
+```
+brew install portaudio
+```
+
+and it will install `portaudio` into `/usr/local/Cellar/portaudio/19.7.0`.
+You need to set the following environment variable
+```
+export PKG_CONFIG_PATH=/usr/local/Cellar/portaudio/19.7.0
+```
+
+so that `pkg-config --cflags --libs portaudio-2.0` can run successfully.
+
+[sherpa-onnx]: https://github.com/k2-fsa/sherpa-onnx

go-api-examples/real-time-speech-recognition-from-microphone/go.mod

@@ -0,0 +1,15 @@
+module real-time-speech-recognition-from-microphone
+
+go 1.20
+
+require (
+	github.com/gordonklaus/portaudio v0.0.0-20230709114228-aafa478834f5
+	github.com/k2-fsa/sherpa-onnx-go v1.5.3-alpha.8
+	github.com/spf13/pflag v1.0.5
+)
+
+require (
+	github.com/k2-fsa/sherpa-onnx-go-linux v1.5.3-alpha.6 // indirect
+	github.com/k2-fsa/sherpa-onnx-go-macos v1.5.3-alpha.4 // indirect
+	github.com/k2-fsa/sherpa-onnx-go-windows v1.5.3-alpha.5 // indirect
+)

go-api-examples/real-time-speech-recognition-from-microphone/go.sum

@@ -0,0 +1,12 @@
+github.com/gordonklaus/portaudio v0.0.0-20230709114228-aafa478834f5 h1:5AlozfqaVjGYGhms2OsdUyfdJME76E6rx5MdGpjzZpc=
+github.com/gordonklaus/portaudio v0.0.0-20230709114228-aafa478834f5/go.mod h1:WY8R6YKlI2ZI3UyzFk7P6yGSuS+hFwNtEzrexRyD7Es=
+github.com/k2-fsa/sherpa-onnx-go v1.5.3-alpha.8 h1:BXc31pWwd7CJLM6m9HavxqiyYJdN3Jc9I26pd4x+JHE=
+github.com/k2-fsa/sherpa-onnx-go v1.5.3-alpha.8/go.mod h1:kszL/pwg9XTpRGi1AYW/aSwdhRBqb6LN0SjXR0jnsBo=
+github.com/k2-fsa/sherpa-onnx-go-linux v1.5.3-alpha.6 h1:o0+l4Wr3IWkWH+kdt8ZZP55L8mRSpW7h1KzvXcfx9FM=
+github.com/k2-fsa/sherpa-onnx-go-linux v1.5.3-alpha.6/go.mod h1:9kU02PSdDdzBApwIDBmE2jWS54WvZbRafOEvW/PVLxE=
+github.com/k2-fsa/sherpa-onnx-go-macos v1.5.3-alpha.4 h1:Xri23R3+tQFkNwO6zVxqZbjMRJP40z7JtoSqW6UP6sM=
+github.com/k2-fsa/sherpa-onnx-go-macos v1.5.3-alpha.4/go.mod h1:a+AJZKNQkFO+JyzGkHySysYfBzzdcoJI5ITFsnhVcmo=
+github.com/k2-fsa/sherpa-onnx-go-windows v1.5.3-alpha.5 h1:OR1LLoptR8W35j5u06KrYUblG35B83HHCPWrVK31uHY=
+github.com/k2-fsa/sherpa-onnx-go-windows v1.5.3-alpha.5/go.mod h1:ermMOETZUv0nM7MmnXWqeHREMR6zQXBhJpEP2fbHIZo=
+github.com/spf13/pflag v1.0.5 h1:iy+VFUOCP1a+8yFto/drg2CJ5u0yRoB7fZw3DKv/JXA=
+github.com/spf13/pflag v1.0.5/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=

go-api-examples/real-time-speech-recognition-from-microphone/main.go

@@ -0,0 +1,110 @@
+package main
+
+import (
+	"github.com/gordonklaus/portaudio"
+	sherpa "github.com/k2-fsa/sherpa-onnx-go/sherpa_onnx"
+	flag "github.com/spf13/pflag"
+	"strings"
+
+	"fmt"
+	"log"
+)
+
+func main() {
+	err := portaudio.Initialize()
+	if err != nil {
+		log.Fatalf("Unable to initialize portaudio: %v\n", err)
+	}
+	defer portaudio.Terminate()
+
+	default_device, err := portaudio.DefaultInputDevice()
+	if err != nil {
+		log.Fatal("Failed to get default input device: %v\n", err)
+	}
+	fmt.Printf("Select default input device: %s\n", default_device.Name)
+	param := portaudio.StreamParameters{}
+	param.Input.Device = default_device
+	param.Input.Channels = 1
+	param.Input.Latency = default_device.DefaultLowInputLatency
+
+	param.SampleRate = 16000
+	param.FramesPerBuffer = 0
+	param.Flags = portaudio.ClipOff
+
+	config := sherpa.OnlineRecognizerConfig{}
+	config.FeatConfig = sherpa.FeatureConfig{SampleRate: 16000, FeatureDim: 80}
+
+	flag.StringVar(&config.ModelConfig.Encoder, "encoder", "", "Path to the encoder model")
+	flag.StringVar(&config.ModelConfig.Decoder, "decoder", "", "Path to the decoder model")
+	flag.StringVar(&config.ModelConfig.Joiner, "joiner", "", "Path to the joiner model")
+	flag.StringVar(&config.ModelConfig.Tokens, "tokens", "", "Path to the tokens file")
+	flag.IntVar(&config.ModelConfig.NumThreads, "num-threads", 1, "Number of threads for computing")
+	flag.IntVar(&config.ModelConfig.Debug, "debug", 0, "Whether to show debug message")
+	flag.StringVar(&config.ModelConfig.ModelType, "model-type", "", "Optional. Used for loading the model in a faster way")
+	flag.StringVar(&config.ModelConfig.Provider, "provider", "cpu", "Provider to use")
+	flag.StringVar(&config.DecodingMethod, "decoding-method", "greedy_search", "Decoding method. Possible values: greedy_search, modified_beam_search")
+	flag.IntVar(&config.MaxActivePaths, "max-active-paths", 4, "Used only when --decoding-method is modified_beam_search")
+	flag.IntVar(&config.EnableEndpoint, "enable-endpoint", 1, "Whether to enable endpoint")
+	flag.Float32Var(&config.Rule1MinTrailingSilence, "rule1-min-trailing-silence", 2.4, "Threshold for rule1")
+	flag.Float32Var(&config.Rule2MinTrailingSilence, "rule2-min-trailing-silence", 1.2, "Threshold for rule2")
+	flag.Float32Var(&config.Rule3MinUtteranceLength, "rule3-min-utterance-length", 20, "Threshold for rule3")
+
+	flag.Parse()
+
+	log.Println("Initializing recognizer (may take several seconds)")
+	recognizer := sherpa.NewOnlineRecognizer(&config)
+	log.Println("Recognizer created!")
+	defer sherpa.DeleteOnlineRecognizer(recognizer)
+
+	stream := sherpa.NewOnlineStream(recognizer)
+
+	// you can choose another value for 0.1 if you want
+	samplesPerCall := int32(param.SampleRate * 0.1) // 0.1 second
+
+	samples := make([]float32, samplesPerCall)
+	s, err := portaudio.OpenStream(param, samples)
+	if err != nil {
+		log.Fatalf("Failed to open the stream")
+	}
+	defer s.Close()
+	chk(s.Start())
+
+	var last_text string
+
+	segment_idx := 0
+
+	fmt.Println("Started! Please speak")
+
+	for {
+		chk(s.Read())
+		stream.AcceptWaveform(int(param.SampleRate), samples)
+
+		for recognizer.IsReady(stream) {
+			recognizer.Decode(stream)
+		}
+
+		text := recognizer.GetResult(stream).Text
+		if len(text) != 0 && last_text != text {
+			last_text = strings.ToLower(text)
+			fmt.Printf("\r%d: %s", segment_idx, last_text)
+		}
+
+		if recognizer.IsEndpoint(stream) {
+			if len(text) != 0 {
+				segment_idx++
+				fmt.Println()
+			}
+			recognizer.Reset(stream)
+		}
+	}
+
+	chk(s.Stop())
+	return
+
+}
+
+func chk(err error) {
+	if err != nil {
+		panic(err)
+	}
+}

go-api-examples/real-time-speech-recognition-from-microphone/run.sh

@@ -0,0 +1,15 @@
+#!/usr/bin/env bash
+
+# Please refer to
+# https://k2-fsa.github.io/sherpa/onnx/pretrained_models/online-transducer/zipformer-transducer-models.html#pkufool-icefall-asr-zipformer-streaming-wenetspeech-20230615-chinese
+# to download the model
+# before you run this script.
+#
+# You can switch to different online models if you need
+
+./real-time-speech-recognition-from-microphone \
+  --encoder ./icefall-asr-zipformer-streaming-wenetspeech-20230615/exp/encoder-epoch-12-avg-4-chunk-16-left-128.onnx \
+  --decoder ./icefall-asr-zipformer-streaming-wenetspeech-20230615/exp/decoder-epoch-12-avg-4-chunk-16-left-128.onnx \
+  --joiner ./icefall-asr-zipformer-streaming-wenetspeech-20230615/exp/joiner-epoch-12-avg-4-chunk-16-left-128.onnx \
+  --tokens ./icefall-asr-zipformer-streaming-wenetspeech-20230615/data/lang_char/tokens.txt \
+  --model-type zipformer2