microsoft/VibeVoice · error · ValueError

No valid entries found in JSON file

Error message

No valid entries found in JSON file

What it means

After the JSON file passes the 'is a list' check, each entry is validated: non-dicts are skipped with a warning, entries missing 'speaker' or 'text' are skipped, non-integer speaker ids are skipped, and empty/whitespace-only text is dropped. If every entry is filtered out, script_lines is empty and this ValueError fires. The earlier warnings in the log tell you which filter killed the entries.

Solutions

1. Check the logger.warning lines emitted just before the exception — they name each skipped entry and why.
2. Ensure every entry is a dict with 'speaker' as an integer-parseable value and a non-empty 'text'.
3. Rename mismatched keys (spk->speaker, transcript->text) with a small jq/python pass before calling the processor.

Example fix

# before
[{"spk": "1", "transcript": "Hi"}]

# after
[{"speaker": "1", "text": "Hi"}]

import json
data = json.load(open(path))
valid = [{'speaker': int(e['speaker']), 'text': e['text'].strip()}
         for e in data
         if isinstance(e, dict) and e.get('speaker') and e.get('text', '').strip()
         and str(e['speaker']).isdigit()]
if not valid:
    raise ValueError('transcript has no usable entries — check speaker/text keys')

def is_valid_entry(e) -> bool:
    return (isinstance(e, dict) and str(e.get('speaker', '')).isdigit()
            and bool(str(e.get('text', '')).strip()))

try:
    enc = processor(text=json_path)
except ValueError as e:
    if 'No valid entries' in str(e):
        logger.error('transcript schema mismatch; expected speaker:int + text:str entries')
        raise
    raise

When it happens

Trigger: A JSON list where all entries have speaker ids like 'speaker_1' or 'one' (int() fails), or all texts are empty strings, or all entries use different keys ('spk'/'utterance' instead of 'speaker'/'text').

Common situations: Schema drift: annotation tools renaming fields; speaker names instead of numeric ids ('narrator'); transcripts where text is stored under 'transcript' rather than 'text'.

Related errors

• JSON file must contain a list of speaker entries
• No valid content found in text file
• Audio at index is too short to be represented
• Audio input is required
• Audio input is required for ASR processing

AI-assisted analysis of microsoft/VibeVoice@94da20d98b (2026-08-15). Data as JSON: /api/errors/71ea277bf5a03b35. Report an issue: GitHub.

Appendix: source

Thrown at vibevoice/processor/vibevoice_processor.py:554

            
            if speaker is None or text is None:
                logger.warning(f"Skipping entry missing speaker or text: {item}")
                continue
            
            # Ensure speaker ID is valid
            try:
                speaker_id = int(speaker)
            except (ValueError, TypeError):
                logger.warning(f"Invalid speaker ID: {speaker}, skipping entry")
                continue
            
            # Clean up text
            text = text.strip()
            if text:
                script_lines.append(f"Speaker {speaker_id}: {text}")
        
        if not script_lines:
            raise ValueError("No valid entries found in JSON file")
            
        return "\n".join(script_lines)

    def _convert_text_to_script(self, text_file: str) -> str:
        """
        Convert text file to script format.
        Handles multiple formats:
        1. Already formatted as "Speaker X: text"
        2. Plain text (assigns to Speaker 1)
        
        Handles edge cases like multiple colons in a line.
        """
        with open(text_file, 'r', encoding='utf-8') as f:
            lines = f.readlines()
        
        script_lines = []
        current_speaker = 1
        

View on GitHub (pinned to 94da20d98b)