voskcli/voskcli.go

package main

import (
	"fmt"
	"io"
	"os"
	"os/exec"
	"os/signal"
	"syscall"
	njson "github.com/m7shapan/njson"
	vosk "github.com/alphacep/vosk-api/go"
)

func usage() {
	fmt.Println(`Usage: voskcli [FILE...]
	
--mic=NAME       Specify the audio device (use 'arecord -L' to find mic name).
--model=PATH     Path to the model to use`)
}

func fatal(a ...any) {
	fmt.Fprintln(os.Stderr, "voskcli:", fmt.Sprint(a...))
	os.Exit(1)
}
func warn(a ...any) {
	fmt.Fprintln(os.Stderr, "voskcli: WARNING:", fmt.Sprint(a...))
}

func print(a ...any) {
	fmt.Fprintln(os.Stdout, fmt.Sprint(a...))
}


func parseResults(json string) string {
	type ResultJson struct {
		Text string `njson:"text"`
		Words []string `njson:"result.#.word"`
		Confs []float64 `njson:"result.#.conf"`
		Starts []float64 `njson:"result.#.start"`
		Ends []float64 `njson:"result.#.end"`
		Confidence float64 `njson:"confidence"`  // only with alternatives
	}
	var s struct {
		Alternatives []ResultJson `njson:"alternatives"`

		// copy paste of ResultJson
		Text string `njson:"text"`
		Words []string `njson:"result.#.word"`
		Confs []float64 `njson:"result.#.conf"`
		Starts []float64 `njson:"result.#.start"`
		Ends []float64 `njson:"result.#.end"`
		Confidence float64 `njson:"confidence"`  // only with alternatives

		ParText string `njson:"partial"`
		ParWords []string `njson:"partial_result.#.word"`
		ParConfs []float64 `njson:"partial_result.#.conf"`
		ParStarts []float64 `njson:"partial_result.#.start"`
		ParEnds []float64 `njson:"partial_result.#.end"`
	}
	err := njson.Unmarshal([]byte(json), &s)
	if err != nil {
		panic(err)
	}
	print(json);
	return s.Text
}


func getMic(mic string) string {
	if mic == "" {
		mic = "default"
	}
	return mic
}

func record(mic string) (io.Reader, error) {
	cmd := exec.Command("arecord", "-q", "-fS16_LE", "-c1", "-r16000", "-D", mic)
	cmd.Stderr = os.Stderr
	stdout, err := cmd.StdoutPipe()
	if err != nil {
		return nil, err
	}
	err = cmd.Start()
	if err != nil {
		return nil, err
	}
	return stdout, nil
}

type Action struct {
	Tags []string
	Text string
}

func main() {
	var opts struct {
		Mic string
		Model string
		Verbose bool
	}
	var model *vosk.VoskModel
	{
		m := os.Getenv("VOSK_MODEL")
		if m == "" {
			fatal("you need to install the a vosk model and set $VOSK_MODEL")
		}
		if opts.Verbose {
			fmt.Fprintln(os.Stderr, "Model: " + m)
		}

		var err error
		model, err = vosk.NewModel(m)
		if err != nil {
			fatal(err)
		}
	}
	defer model.Free()

	var r *vosk.VoskRecognizer
	{
		var err error
		r, err = vosk.NewRecognizer(model, float64(16000))
		if err != nil {
			panic(err)
		}
		r.SetWords(1)
		r.SetMaxAlternatives(3)//3 ? 10 ?
	}
	defer r.Free()

	var mic string
	var audio io.Reader

	mic = getMic(opts.Mic)
	if opts.Verbose {
		fmt.Fprintln(os.Stderr, "Microphone: " + mic)
	}
	var err error
	audio, err = record(mic)
	if err != nil {
		fatal(err)
	}

	terminate := make(chan os.Signal, 1)
	signal.Notify(terminate, os.Interrupt, syscall.SIGHUP, syscall.SIGINT, syscall.SIGTERM)

	for {
		select {
			case <-terminate: return
			default:
		}
		chunk := make([]byte, 4096)
		_, err := io.ReadFull(audio, chunk)
		if err != nil {
			if err != io.EOF && err != io.ErrUnexpectedEOF {
				panic(err)
			}
			r, err := record(mic)
			if err != nil {
				warn(err)
			}
			audio = r
			continue
		}

		var res string
		if  (r.AcceptWaveform(chunk) == 1) {
			res = r.FinalResult()
			parseResults(res);
		} else {
			res = r.PartialResult() // r.Result()
			parseResults(res);
		}
	}

	
}