When developing games or apps individually, there are tasks that take more time than coding. That is, "BGM and Sound Effect Selection".

Spending hours browsing free audio sites without finding a track that matches your imagination, and not having the budget to hire a professional composer... Many indie developers face such challenges, don't they?

Recently, AI tools that can generate music from text have increased, but this time I will share an AI music generation approach that can be practically integrated into the workflow of developers and creators, along with the verification results.

Challenges with Traditional AI Music Tools

Currently, many Text-to-Music (Text-based Music Generation) AI tools exist. However, the following challenges were encountered when integrating them into game development.

• Only outputs monolithic audio files (MP3/WAV)
• Cannot make micro-adjustments for specific scenes within the game, such as "muting drums only" or "lowering the bass volume"
• Cannot edit chord progressions or melody lines in DAWs (music production software)

In other words, the main issue was that they were only output as finished products, leaving little room for engineers and creators to make adjustments later (lack of extensibility).

Verified Tool: Muzegen AI

The tool we verified to address these challenges is the creator-focused AI music generation platform Muzegen AI.

In addition to the basic function of generating high-quality music from prompts (text instructions), it also implements "Functions that allow for post-editing", which is very beneficial for developers.

Benefits from the Engineer/Creator Perspective

The biggest advantage of introducing Muzegen AI into your workflow is the following two points.

1. Flexible Audio Control via Stem (Multi-track) Separation

You can export the generated music as stems (separate audio tracks) such as vocals, drums, bass, and other instruments.

• Use cases in development:
  • Controlling the volume of each track individually on Unity and Unreal Engine.
  • Dynamic BGM effects (interactive music), such as fading in the drum track during the latter half of a boss battle, can now be realized even with music generated by AI.

2. MIDI Export Functionality for Reconstruction

A feature that allows output as MIDI data (performance information of melody and chords) rather than just waveform data.

• Use cases in development:
  • Load the exported MIDI into your DAW on your PC and replace the sound source (e.g., Velocity) with your preferred one.
  • Correct the last note of the generated melody to align with the program.

3. Commercial Use Compliance (Royalty-free)

When selling indie games on Steam or mobile app stores, the ability to incorporate BGM without worrying about copyright infringement also contributes to reducing legal work hours.

Summary

The process has evolved from 'having AI create a song' to 'decomposing the AI-generated materials as parts and integrating them into your project for optimization'. AI-generated materials as parts and integrating them into your project for optimization .

Developers who often get stuck in BGM searching should be able to drastically reduce audio production work (by about 90%) by utilizing generators that support Stem/MIDI output, such as Muzegen AI.

As an option to focus limited personal development resources on core logic implementation and enhancing the game's fun, please definitely try using such generators.