Music sets the tone for video content, especially product demos and explainer videos.
It cues the audience on how to feel, and when the content is synchronized to the music, it adds a level of polish that separates premium videos from basic, run-of-the mill content.
We’ve sourced and licensed hundreds of tracks, and it can be a painstaking process full of friction and dead ends. We’ve refined our approach using AI to make it repeatable, and we spell out our process below.
Process Overview
Here is the 4-step process:
- Find reference videos
- Use ChatGPT to extract the “music DNA”
- Convert this into a music brief and search terms
- Use the brief to source tracks across multiple music licensing platforms
Beginning with Step 1, identify 4-6 reference videos with music that matches the tone, pace, and energy that you envision for your video. These videos don’t need to be exactly what you’re aiming for, just directionally aligned. Remember the adage: garbage in, garbage out. The quality of your reference videos affects the quality of the results.
Take the video references (ideally from YouTube or Vimeo, X links won’t work as well) and enter them into ChatGPT. Use this prompt below.
NOTE: We’ve tested this across multiple LLMs, but ChatGPT consistently produces the most usable output.
ChatGPT Prompt
I’m a video producer sourcing music for client work. I’m going to share several video links.
For each video, analyze the background music and provide a detailed breakdown with the following:
1. Music Profile
- Genre and subgenre
- Tempo (approximate BPM range)
- Instrumentation (specific elements like synths, piano, pulses, textures, etc.)
- Structure (how the track evolves over time)
2. Creative / Emotional Direction
- Emotional tone (e.g., tension, optimism, introspection)
- What kind of brand or message this music supports
- Where it would be used (e.g., product demo, founder story, launch video)
3. Music Brief
- Write a concise, 1–2 sentence brief that I could reuse to source similar tracks.
4. Sourcing Guidance
- Provide high-quality search terms for:
- APM Music
- Musicbed
- Artlist
- Envato
- AudioJungle
- Suggest specific artists or styles to explore on those platforms
5. Similarity Mapping
- Explain what makes this track effective and what NOT to choose (what would feel too corporate, too hype, etc.)
Important:
- Prioritize cinematic, tasteful, non-corporate music (avoid generic “corporate upbeat” recommendations)
- Think like a high-end creative agency, not a stock music library
- Focus on helping me replicate the feel, not just identify the exact song
Here are the video links:
[PASTE YOUR VIDEO LINKS]
How to use the ChatGPT output
Here’s the step-by-step process for how to use the output.
1. Start with the music brief
Use the 1-2 sentence brief as your primary search input across each of your preferred music licensing sites.
2. Use search terms and artists to refine
Layer in the recommended keywords to narrow the results. Or search specific artist recommendations.
3. Iterate with feedback
Incorporate additional information into your ChatGPT prompt based on your search process:
- Add tracks you like and refine toward them (even if they’re not perfect matches)
- Add tracks you don’t like and filter them out (and be specific regarding what you don’t like)
These signals help to refine the output.
4. Leverage other outputs
The music profile and creative direction are useful in case you need to articulate why you chose specific songs to a colleague or client.
5. Song matches
For any tracks that exactly match your preferences, add them to ChatGPT to build a feedback loop, which gives you more control and better results. Even though each music library has built-in “similar track” features, they’re often inconsistent.
Want to discuss your next video project? Schedule a free discovery call here or click below.
