refactor: add vav header

service/groq_stt.go

@@ -3,6 +3,7 @@ package service
 import (
 	"bytes"
 	"context"
+	"encoding/binary"
 	"encoding/json"
 	"fmt"
 	"io"
@@ -26,23 +27,51 @@ func (g *GroqSTT) TranscribeChunk(ctx context.Context, audio []byte) (string, er
 	return g.callGroqAPI(ctx, audio)
 }
 
-func (g *GroqSTT) callGroqAPI(ctx context.Context, audioData []byte) (string, error) {
+func addWavHeader(pcmData []byte) []byte {
+	const sampleRate = 16000
+	const numChannels = 1
+	const bitsPerSample = 16
+
+	header := new(bytes.Buffer)
+
+	header.WriteString("RIFF")
+	binary.Write(header, binary.LittleEndian, int32(36+len(pcmData)))
+	header.WriteString("WAVE")
+
+	header.WriteString("fmt ")
+	binary.Write(header, binary.LittleEndian, int32(16))
+	binary.Write(header, binary.LittleEndian, int16(1))
+	binary.Write(header, binary.LittleEndian, int16(numChannels))
+	binary.Write(header, binary.LittleEndian, int32(sampleRate))
+	binary.Write(header, binary.LittleEndian, int32(sampleRate*numChannels*bitsPerSample/8))
+	binary.Write(header, binary.LittleEndian, int16(numChannels*bitsPerSample/8))
+	binary.Write(header, binary.LittleEndian, int16(bitsPerSample))
+
+	header.WriteString("data")
+	binary.Write(header, binary.LittleEndian, int32(len(pcmData)))
+
+	return append(header.Bytes(), pcmData...)
+}
+
+func (g *GroqSTT) callGroqAPI(ctx context.Context, rawAudio []byte) (string, error) {
 	if g.apiKey == "" {
 		return "", fmt.Errorf("API Key is empty")
 	}
 
+	wavData := addWavHeader(rawAudio)
+
 	body := &bytes.Buffer{}
 	writer := multipart.NewWriter(body)
 
 	part, err := writer.CreateFormFile("file", "audio.wav")
 	if err != nil {
-		return "", fmt.Errorf("create from file failed: %w", err)
+		return "", fmt.Errorf("create form file failed: %w", err)
 	}
-	part.Write(audioData)
+	part.Write(wavData)
 
 	writer.WriteField("model", "whisper-large-v3")
 	writer.WriteField("response_format", "json")
-	//writer.WriteField("language", "id")
+	writer.WriteField("language", "id")
 	writer.Close()
 
 	url := "https://api.groq.com/openai/v1/audio/transcriptions"