Go + FFmpeg: How to Segment (Chunk) Audio File

segment_audio.go

type File struct {
	Name        string
	Start       time.Duration
	End         time.Duration
	ContentType string
}

func SegmentAudio(ctx context.Context, filename string, maxDuration time.Duration, dir string, pattern string) ([]*File, error) {
	if dir == "" {
		dir = os.TempDir()
	}
	if pattern == "" {
		pattern = "%d"
	}
	if strings.Contains(fmt.Sprintf(pattern, 0), "%!") {
		return nil, fmt.Errorf("invalid pattern: %v", pattern)
	}
	formatName, err := FormatName(ctx, filename)
	if err != nil {
		return nil, err
	}
	segmentTime := strconv.FormatFloat(maxDuration.Seconds(), 'f', -1, 64)
	f, err := os.CreateTemp("", "")
	if err != nil {
		return nil, err
	}
	defer os.Remove(f.Name())
	defer f.Close()
	ctx, cancel := context.WithCancel(ctx)
	defer cancel()
	cmd := exec.CommandContext(
		ctx,
		"ffmpeg",
		"-f", formatName,
		"-i", filename,
		"-codec", "copy",
		"-map", "0",
		"-f", "segment",
		"-segment_list_type", "csv",
		"-segment_list", f.Name(),
		"-segment_time", segmentTime,
		"-segment_format", formatName,
		filepath.Join(dir, pattern),
	)
	stderr, err := cmd.StderrPipe()
	if err != nil {
		return nil, err
	}
	if err := cmd.Start(); err != nil {
		return nil, err
	}
	slurp, _ := io.ReadAll(stderr)
	slog.Debug("ffmpeg", slog.String("stderr", string(slurp)))
	if err := cmd.Wait(); err != nil {
		return nil, err
	}
	var files []*File
	csvReader := csv.NewReader(f)
	for {
		record, err := csvReader.Read()
		if err == io.EOF {
			break
		}
		if err != nil {
			return nil, err
		}
		if len(record) != 3 {
			return nil, fmt.Errorf("invalid segment list: %v", record)
		}
		fname := filepath.Join(dir, record[0])
		f, err := os.Open(fname)
		if err != nil {
			return nil, err
		}
		defer f.Close()
		contentType, err := detectContentType(f)
		if err != nil {
			return nil, err
		}
		start, err := time.ParseDuration(record[1] + "s")
		if err != nil {
			return nil, err
		}
		end, err := time.ParseDuration(record[2] + "s")
		if err != nil {
			return nil, err
		}
		files = append(files, &File{Name: fname, Start: start, End: end, ContentType: contentType})
	}
	return files, nil
}

func FormatName(ctx context.Context, filename string) (string, error) {
	ctx, cancel := context.WithCancel(ctx)
	defer cancel()
	cmd := exec.CommandContext(ctx, "ffprobe", "-i", filename, "-show_format", "-output_format", "json")
	stdout, err := cmd.StdoutPipe()
	if err != nil {
		return "", nil
	}
	if err := cmd.Start(); err != nil {
		return "", nil
	}
	var output struct {
		Format struct {
			FormatName string `json:"format_name"`
			Duration   string `json:"duration"`
		} `json:"format"`
	}
	if err := json.NewDecoder(stdout).Decode(&output); err != nil {
		return "", nil
	}
	if err := cmd.Wait(); err != nil {
		return "", nil
	}
	return output.Format.FormatName, nil
}