The Hi-Audio online platform for recording and distributing multi-track music datasets.

Area of Science:

• Music Technology
• Computer Science
• Human-Computer Interaction

Background:

• Music Information Retrieval (MIR) research requires accessible, open-access audio datasets.
• Current tools for collaborative music creation and research lack integrated analysis and latency estimation.

Purpose of the Study:

• Introduce the Hi-Audio platform, an open-source tool for recording, uploading, and sharing multitrack musical compositions.
• Advance Music Information Retrieval (MIR) research through an open-access audio database.
• Provide a robust method for round-trip latency estimation in browser-based audio applications.

Main Methods:

• Developed a web-based platform (hiaudio.fr) for asynchronous collaboration and audio sharing.
• Implemented automatic audio analysis using signal processing and machine learning for metadata generation.
• Integrated a novel browser-based round-trip latency estimation technique.

Main Results:

• The Hi-Audio platform successfully supports recording, uploading, and sharing of multitrack compositions.
• Automatic analysis generates rich metadata for uploaded audio files.
• User evaluation demonstrated high task completion rates and effective usability for musicians with minimal training.

Conclusions:

• The Hi-Audio platform effectively facilitates remote, asynchronous collaboration for musicians and researchers.
• The integrated latency estimation method is applicable to real-world distributed recording scenarios.
• The platform advances open-access audio data collection for Music Information Retrieval (MIR) research.