We would like to add/support STT/TTS from other providers though, if that’s what you mean. If you end up building support for that, we accept PRs! 😊

Hi Bro! I have an question about KITT.Currently, I’m trying to read out the speech stream as you did in

lkgpt-service/pkg/service/transcriber.go

, since the ASR API I tried to use just received PCM raw data, so I tried to modify the code in

start()

to use https://github.com/hraban/opus to decode ogg data to pcm data as following

stream, err:= opus.NewStream(t.oggReader)
                        if err != nil {
                                logger.Errorw("failed to create opus decoder", err)
                        }

but got

failed to create opus decoder	{"error": "OP_EBADTIMESTAMP"}

error. I’m new to golang, so have no idea if I use a wrong lib? or any other ways to get pcm raw data ? THANK YOU VERY MUCH!

Hi Team, I added a function by using “mccoy.space/g/ogg”

import (	
   "<http://mccoy.space/g/ogg|mccoy.space/g/ogg>"
)

func OggToPCM(data []byte) ([]byte) {
        decoder := ogg.NewDecoder(bytes.NewReader(data))
        page, err := decoder.Decode()
        if err != nil {
                logger.Errorw("Error while decoding opus file: %v", err)
                return nil
        }
        //fmt.Println("ogg data has %d pages", len(page.Packets))
        audiodata := make([]byte, 0)
        for i:=0; i< len(page.Packets); i++ {
                audiodata = append(audiodata, page.Packets[i]...)
        }
        return audiodata
}

and update

transcriber.go

as following

go func() {
                        defer close(nextCh)
                        buf := make([]byte, 1024)
                        for {
                                select {
                                case <-endStreamCh:
                                        return
                                default:
                                        n, err := t.oggReader.Read(buf)
                                        if err != nil {
                                                if err != io.EOF {
                                                        logger.Errorw("failed to read from ogg reader", err)
                                                }
                                                return
                                        }

                                        if n <= 0 {
                                                fmt.Printf("No Data read from OggReader \n")
                                                continue // No data
                                        }

                                        opusdata := OggToPCM(buf[:n])
                                        if opusdata != nil {
                                                opusbytes = append(opusbytes, opusdata...)
                                        } else {
                                                fmt.Printf("No Data from OpusDecode \n")
                                        }
                                }
                        }
                }()

but seems it’s not working neither, any ideas on handle the buffer from

oggReader

and convert it to PCM raw data?