反谱AI

What

反谱 AI is a web-based music processing product focused on turning existing music materials into editable score data. Based on the page, it serves music students, hobbyists, and musicians who need to recognize sheet music from images or PDFs, transcribe audio into staff notation or numbered notation, separate vocals from accompaniment, and convert MIDI or score files between common music notation formats.

Its core workflow is built around uploading source material, running an AI-assisted processing task, previewing the result online, and then downloading or further editing the output. The product appears positioned as a practical utility for score digitization, transcription, and basic music production support rather than a full digital audio workstation or full-featured notation platform.

Features

• Image and PDF score recognition: Converts staff notation images or PDF files into machine-readable music notation such as MusicXML, with support for generating numbered notation.
• Audio-to-score transcription: Transcribes music audio into staff notation and numbered notation, helping users turn recordings into readable musical structure.
• Vocal and accompaniment separation: Splits songs into vocal and instrumental parts for practice, arrangement, or remix preparation.
• Score and notation file conversion: Converts MIDI and supported score formats into staff notation and numbered notation for easier reuse in notation software.
• Online MIDI import and editing: Provides browser-based MIDI import and real-time score handling, which is useful for quick adjustments without installing desktop software.
• Batch processing and online preview: Higher-tier memberships support batch uploads, while the platform also offers online preview and download workflows for processed results.

Helpful Tips

• For OCR-style score recognition, input quality matters; clean, high-resolution sheet images or well-structured PDFs will likely improve usable results.
• For audio transcription, treat the output as a draft for review, especially on dense polyphonic music, expressive performances, or noisy recordings, since the page says accuracy is still being improved.
• Check file-size, duration, and retention limits before rollout in a classroom, studio, or archive workflow, because capabilities differ significantly between free and paid tiers.
• If your work depends on reusable notation files, prioritize formats like MusicXML and MIDI that can be edited in notation software such as MuseScore.
• The platform uses a token-like “音符” system for tasks and unlocking content, so teams should estimate usage volume in advance to avoid friction in repeated processing.

OpenClaw Skills

Within the OpenClaw ecosystem, 反谱 AI could likely support skills for music-ingestion pipelines: an agent could accept image, PDF, audio, MIDI, or supported music links, route each input to the right processing mode, and then organize outputs into MusicXML, MIDI, numbered notation, or separated stems. If native integration is not provided by the product page, this should be treated as a likely workflow design rather than a confirmed built-in connection.

This combination could enable useful profession-specific agents for music educators, arrangers, choir directors, and content libraries. Likely examples include an “archive-to-score” workflow that digitizes old printed scores, a “rehearsal prep” agent that separates accompaniment and produces simplified notation, or a “music research assistant” that indexes uploaded materials by title, source type, and output format. In practice, that could reduce manual transcription overhead and make music materials more searchable, editable, and reusable across teaching and production environments.