feat: initial commit

internal/infrastructure/speech/audio_converter.go

@@ -0,0 +1,63 @@
+package speech
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+	"os"
+	"os/exec"
+	"path/filepath"
+
+	"github.com/google/uuid"
+)
+
+type FFmpegConverter struct {
+	ffmpegPath string
+	tempDir    string
+}
+
+func NewFFmpegConverter(ffmpegPath, tempDir string) *FFmpegConverter {
+	if ffmpegPath == "" {
+		ffmpegPath = "ffmpeg"
+	}
+	if tempDir == "" {
+		tempDir = os.TempDir()
+	}
+	return &FFmpegConverter{ffmpegPath: ffmpegPath, tempDir: tempDir}
+}
+
+func (c *FFmpegConverter) Convert(ctx context.Context, input []byte, fromMime, toMime string) ([]byte, error) {
+	id := uuid.New().String()
+	inFile := filepath.Join(c.tempDir, id+".input")
+	outFile := filepath.Join(c.tempDir, id+".wav")
+
+	defer os.Remove(inFile)
+	defer os.Remove(outFile)
+
+	if err := os.WriteFile(inFile, input, 0600); err != nil {
+		return nil, fmt.Errorf("write temp input: %w", err)
+	}
+
+	cmd := exec.CommandContext(ctx, c.ffmpegPath,
+		"-i", inFile,
+		"-ar", "16000",
+		"-ac", "1",
+		"-f", "wav",
+		"-y",
+		outFile,
+	)
+
+	var stderr bytes.Buffer
+	cmd.Stderr = &stderr
+
+	if err := cmd.Run(); err != nil {
+		return nil, fmt.Errorf("ffmpeg conversion: %w: %s", err, stderr.String())
+	}
+
+	out, err := os.ReadFile(outFile)
+	if err != nil {
+		return nil, fmt.Errorf("read converted file: %w", err)
+	}
+
+	return out, nil
+}

internal/infrastructure/speech/openai_whisper.go

@@ -0,0 +1,89 @@
+package speech
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"mime/multipart"
+	"net/http"
+	"time"
+)
+
+const whisperAPIURL = "https://api.openai.com/v1/audio/transcriptions"
+
+type OpenAIWhisper struct {
+	apiKey   string
+	model    string
+	language string
+	client   *http.Client
+}
+
+func NewOpenAIWhisper(apiKey, model, language string) *OpenAIWhisper {
+	return &OpenAIWhisper{
+		apiKey:   apiKey,
+		model:    model,
+		language: language,
+		client:   &http.Client{Timeout: 60 * time.Second},
+	}
+}
+
+type whisperResponse struct {
+	Text string `json:"text"`
+}
+
+func (w *OpenAIWhisper) Transcribe(ctx context.Context, audioData []byte, mimeType string) (string, error) {
+	var buf bytes.Buffer
+	mw := multipart.NewWriter(&buf)
+
+	fw, err := mw.CreateFormFile("file", "audio.wav")
+	if err != nil {
+		return "", fmt.Errorf("create form file: %w", err)
+	}
+	if _, err := fw.Write(audioData); err != nil {
+		return "", fmt.Errorf("write audio data: %w", err)
+	}
+
+	if err := mw.WriteField("model", w.model); err != nil {
+		return "", fmt.Errorf("write model field: %w", err)
+	}
+	if err := mw.WriteField("response_format", "json"); err != nil {
+		return "", fmt.Errorf("write response_format: %w", err)
+	}
+	if w.language != "" {
+		if err := mw.WriteField("language", w.language); err != nil {
+			return "", fmt.Errorf("write language field: %w", err)
+		}
+	}
+	mw.Close()
+
+	req, err := http.NewRequestWithContext(ctx, http.MethodPost, whisperAPIURL, &buf)
+	if err != nil {
+		return "", fmt.Errorf("create whisper request: %w", err)
+	}
+	req.Header.Set("Authorization", "Bearer "+w.apiKey)
+	req.Header.Set("Content-Type", mw.FormDataContentType())
+
+	resp, err := w.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("whisper API call: %w", err)
+	}
+	defer resp.Body.Close()
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("read whisper response: %w", err)
+	}
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("whisper API error %d: %s", resp.StatusCode, string(body))
+	}
+
+	var result whisperResponse
+	if err := json.Unmarshal(body, &result); err != nil {
+		return "", fmt.Errorf("parse whisper response: %w", err)
+	}
+
+	return result.Text, nil
+}