sipeed/picoclaw · error

failed to stat audio file %s: %w

Error message

failed to stat audio file %s: %w

What it means

Thrown when audioFile.Stat() fails right after a successful os.Open in WhisperTranscriber.Transcribe. Same rare class as the ElevenLabs stat error (303): the fd is open, so fstat fails only on a concurrently-closed handle, a broken network filesystem (stale NFS handle, EIO), or a kernel-level surprise. The size from this Stat feeds later logging and doRequest.

Source

Thrown at pkg/audio/asr/whisper_transcriber.go:148

	return t.doRequest(ctx, &requestBody, writer.FormDataContentType(), int64(len(data)))
}

func (t *WhisperTranscriber) Transcribe(ctx context.Context, audioFilePath string) (*TranscriptionResponse, error) {
	logger.InfoCF("voice", "Starting whisper transcription", map[string]any{
		"audio_file": audioFilePath,
		"model":      t.modelID,
		"provider":   t.providerName,
	})

	audioFile, err := os.Open(audioFilePath)
	if err != nil {
		return nil, fmt.Errorf("failed to open audio file %s: %w", audioFilePath, err)
	}
	defer audioFile.Close()

	fileInfo, err := audioFile.Stat()
	if err != nil {
		return nil, fmt.Errorf("failed to stat audio file %s: %w", audioFilePath, err)
	}

	var requestBody bytes.Buffer
	writer := multipart.NewWriter(&requestBody)

	part, err := writer.CreateFormFile("file", filepath.Base(audioFilePath))
	if err != nil {
		return nil, fmt.Errorf("failed to create form file: %w", err)
	}

	if _, copyErr := io.Copy(part, audioFile); copyErr != nil {
		return nil, fmt.Errorf("failed to copy audio data: %w", copyErr)
	}

	if err = writer.WriteField("model", t.modelID); err != nil {
		return nil, fmt.Errorf("failed to write model field: %w", err)
	}

View on GitHub (pinned to 49183d7e8d)

Solutions

1. Serialize cleanup and transcription (delete only after Transcribe returns).
2. Copy network-mounted audio to local disk first.
3. Run with -race to find concurrent Close of the same path.

// The library Stats right after Open; do the same check first to own the error message:
fi, err := os.Stat(path)
if err != nil { return fmt.Errorf("audio unstable before transcription: %w", err) }
if fi.Size() == 0 { return errors.New("empty audio file") }

func isStatFailure(err error) bool {
    var pe *os.PathError
    return errors.As(err, &pe) && pe.Op == "stat"
}

if _, err := whisper.Transcribe(ctx, path); err != nil {
    if isStatFailure(err) {
        // fd/fs-level inconsistency: serialize lifecycle, copy locally, retry once
        return retryOnceAfterCopy(ctx, path)
    }
    return err
}

When it happens

Trigger: Another goroutine closes or replaces the file between Open and Stat; NFS/FUSE mount returns an error for fstat; fd table corruption after an fd leak elsewhere.

Common situations: Race between temp-file cleanup and transcription start; containerized workloads on network volumes.

Related errors

• failed to get file info: %w
• failed to open audio file %s: %w
• failed to read audio file: %w
• failed to open audio file: %w
• failed to copy file content: %w

AI-assisted analysis of sipeed/picoclaw@49183d7e8d (2026-08-15). Data as JSON: /api/errors/704b244835192609. Report an issue: GitHub.