jamiepine/voicebox · error · ValueError

Audio exceeds {MAX_TRANSCRIBE_BYTES // (1024 * 1024)} MB lim

Error message

Audio exceeds {MAX_TRANSCRIBE_BYTES // (1024 * 1024)} MB limit.

What it means

Raised by voicebox_transcribe in base64 mode when the decoded byte length len(raw) exceeds MAX_TRANSCRIBE_BYTES (200 MB). Unlike error 50 this fires after decoding, so it guards inflated base64 payloads rather than on-disk file size.

Source

Thrown at backend/mcp_server/tools.py:160

                )
            path = Path(audio_path)
            if not path.is_absolute():
                raise ValueError("`audio_path` must be absolute.")
            if not path.is_file():
                raise ValueError(f"File not found: {audio_path}")
            if path.stat().st_size > MAX_TRANSCRIBE_BYTES:
                raise ValueError(
                    f"File exceeds {MAX_TRANSCRIBE_BYTES // (1024 * 1024)} MB limit."
                )
            return await _transcribe_file(path, language, model)

        # Base64 mode: decode into a temp file, transcribe, clean up.
        try:
            raw = b64.b64decode(audio_base64, validate=True)
        except Exception as exc:
            raise ValueError(f"Invalid audio_base64: {exc}") from exc
        if len(raw) > MAX_TRANSCRIBE_BYTES:
            raise ValueError(
                f"Audio exceeds {MAX_TRANSCRIBE_BYTES // (1024 * 1024)} MB limit."
            )
        with tempfile.NamedTemporaryFile(
            suffix=".wav", delete=False
        ) as tmp:
            tmp.write(raw)
            tmp_path = Path(tmp.name)
        try:
            return await _transcribe_file(tmp_path, language, model)
        finally:
            tmp_path.unlink(missing_ok=True)

    @mcp.tool(
        name="voicebox.list_captures",
        description=(
            "List recent voice captures (dictations, recordings, uploads) "
            "with their transcripts. Most-recent first."
        ),

View on GitHub (pinned to 51f49dea19)

Solutions

1. Reduce the decoded audio to under 200 MB (trim, segment, or re-encode to 16 kHz mono).
2. For local clients, switch to audio_path to skip the base64 round trip and the same 200 MB cap on file size.
3. Batch long recordings into multiple under-limit calls.

Example fix

// before
voicebox_transcribe(audio_base64=b64encode(huge_wav))
// after
segments = split(huge_wav, max_bytes=150*1024*1024)
for s in segments: voicebox_transcribe(audio_base64=b64encode(s))

import base64
from backend.mcp_server.tools import MAX_TRANSCRIBE_BYTES
raw = base64.b64decode(audio_base64, validate=True)
if len(raw) > MAX_TRANSCRIBE_BYTES:
    raise ValueError("decoded audio exceeds 200 MB; segment it first")
await voicebox_transcribe(audio_base64=audio_base64)

def decoded_audio_under_limit(value: str, limit: int = 200 * 1024 * 1024) -> bool:
    import base64
    try:
        return len(base64.b64decode(value, validate=True)) <= limit
    except Exception:
        return False

try:
    await voicebox_transcribe(audio_base64=audio_base64)
except ValueError as exc:
    if "exceeds" in str(exc) and "MB limit" in str(exc):
        for chunk in segment_audio_bytes(raw, max_bytes=150 * 1024 * 1024):
            await voicebox_transcribe(audio_base64=base64.b64encode(chunk).decode())
    else:
        raise

When it happens

Trigger: Sending a base64-encoded clip whose decoded content is over 200 MB; base64 of a >150 MB binary will already approach the cap (base64 inflation is ~4/3).

Common situations: Long uncompressed audio shipped via base64; transport ceiling lower than the decode cap so the payload often fails earlier; client assuming the limit applies to the encoded size, not decoded.

Related errors

• File exceeds {MAX_TRANSCRIBE_BYTES // (1024 * 1024)} MB limi
• Invalid audio_base64: {exc}
• Pass exactly one of `audio_base64` or `audio_path`.
• `audio_path` is only available to loopback callers — remote
• `audio_path` must be absolute.

AI-assisted analysis of jamiepine/voicebox@51f49dea19 (2026-08-12). Data as JSON: /api/errors/e9e04daee13e83eb. Report an issue: GitHub.