This repository has been archived on 2025-08-26. You can view files and clone it, but cannot push or open issues or pull requests.
Files
enginex_bi_series-sherpa-onnx/go-api-examples/non-streaming-canary-decode-files/main.go
Fangjun Kuang fd9a687ec2 Add Pascal/Go/C#/Dart API for NeMo Canary ASR models (#2367)
Add support for the new NeMo Canary ASR model across multiple language bindings by introducing a Canary model configuration and setter method on the offline recognizer.

- Define Canary model config in Pascal, Go, C#, Dart and update converter functions
- Add SetConfig API for offline recognizer (Pascal, Go, C#, Dart)
- Extend CI/workflows and example scripts to test non-streaming Canary decoding
2025-07-10 14:53:33 +08:00

114 lines
2.9 KiB
Go

package main
import (
"bytes"
"encoding/binary"
"log"
"os"
"strings"
sherpa "github.com/k2-fsa/sherpa-onnx-go/sherpa_onnx"
"github.com/youpy/go-wav"
)
func main() {
log.SetFlags(log.LstdFlags | log.Lmicroseconds)
config := sherpa.OfflineRecognizerConfig{}
config.ModelConfig.Canary.Encoder = "./sherpa-onnx-nemo-canary-180m-flash-en-es-de-fr-int8/encoder.int8.onnx"
config.ModelConfig.Canary.Decoder = "./sherpa-onnx-nemo-canary-180m-flash-en-es-de-fr-int8/decoder.int8.onnx"
config.ModelConfig.Canary.SrcLang = "en"
config.ModelConfig.Canary.TgtLang = "en"
config.ModelConfig.Canary.UsePnc = 1
config.ModelConfig.Tokens = "./sherpa-onnx-nemo-canary-180m-flash-en-es-de-fr-int8/tokens.txt"
waveFilename := "./sherpa-onnx-nemo-canary-180m-flash-en-es-de-fr-int8/test_wavs/en.wav"
samples, sampleRate := readWave(waveFilename)
log.Println("Initializing recognizer (may take several seconds)")
recognizer := sherpa.NewOfflineRecognizer(&config)
log.Println("Recognizer created!")
defer sherpa.DeleteOfflineRecognizer(recognizer)
log.Println("Start decoding!")
stream := sherpa.NewOfflineStream(recognizer)
defer sherpa.DeleteOfflineStream(stream)
stream.AcceptWaveform(sampleRate, samples)
recognizer.Decode(stream)
log.Println("Decoding done!")
result := stream.GetResult()
log.Println("Text in English: " + strings.ToLower(result.Text))
s := sherpa.NewOfflineStream(recognizer)
defer sherpa.DeleteOfflineStream(s)
s.AcceptWaveform(sampleRate, samples)
config.ModelConfig.Canary.TgtLang = "de"
recognizer.SetConfig(&config)
recognizer.Decode(s)
result = s.GetResult()
log.Println("Text in German: " + strings.ToLower(result.Text))
}
func readWave(filename string) (samples []float32, sampleRate int) {
file, _ := os.Open(filename)
defer file.Close()
reader := wav.NewReader(file)
format, err := reader.Format()
if err != nil {
log.Fatalf("Failed to read wave format")
}
if format.AudioFormat != 1 {
log.Fatalf("Support only PCM format. Given: %v\n", format.AudioFormat)
}
if format.NumChannels != 1 {
log.Fatalf("Support only 1 channel wave file. Given: %v\n", format.NumChannels)
}
if format.BitsPerSample != 16 {
log.Fatalf("Support only 16-bit per sample. Given: %v\n", format.BitsPerSample)
}
reader.Duration() // so that it initializes reader.Size
buf := make([]byte, reader.Size)
n, err := reader.Read(buf)
if n != int(reader.Size) {
log.Fatalf("Failed to read %v bytes. Returned %v bytes\n", reader.Size, n)
}
samples = samplesInt16ToFloat(buf)
sampleRate = int(format.SampleRate)
return
}
func samplesInt16ToFloat(inSamples []byte) []float32 {
numSamples := len(inSamples) / 2
outSamples := make([]float32, numSamples)
for i := 0; i != numSamples; i++ {
s := inSamples[i*2 : (i+1)*2]
var s16 int16
buf := bytes.NewReader(s)
err := binary.Read(buf, binary.LittleEndian, &s16)
if err != nil {
log.Fatal("Failed to parse 16-bit sample")
}
outSamples[i] = float32(s16) / 32768
}
return outSamples
}