How to Pick Royalty-Free Music for Client Videos with Beatoven.ai

November 21, 2026

Beatoven.ai makes getting a track easy. The part that gets agencies into trouble is everything after the download: rights, loudness, and whether the client’s ad account will flag it.

This is the process-side view for people delivering video to clients, not the “how do I generate music” basics.

Why does generated music sometimes still get flagged?

Two reasons. First, some platforms run audio fingerprinting on any music and flag anything unfamiliar, then release the claim once you submit documentation. Second, the licence you hold may not cover the specific use — for example paid social advertising versus organic posting can be treated differently by the platform, not by the music service.

What to keep for every delivery:

  1. The generation record (project, date, duration).
  2. The licence tier you were on when you generated it.
  3. Screenshots of the terms in force at that date.

That file takes five minutes to assemble and answers 90 percent of future claims.

How do you match music to a client’s brand?

Build a mood map once per client, then reuse it. For each client, record: primary mood, intensity ceiling, instrumentation to avoid (no synth brass for a law firm), and two reference tracks they have approved before.

The value is velocity — the third video for a client takes ten minutes because the decisions are already made. Also, consistency: a channel where every video sounds like it comes from the same studio reads as a brand, not a montage.

What loudness should you target?

The music must sit under dialogue. Practical targets vary by platform, but a common working range is dialogue peaking around -6 dB and the music bed 15 to 20 dB below dialogue during speech.

Generated tracks arrive mastered and often loud. Turn them down; do not ask the generator for a quieter track. And duck the bed under speech with a sidechain if your editor supports it — a flat 15 dB drop across the whole video leaves gaps in the quiet parts.

How do you handle revisions without regenerating?

The client says “can we try something different” about the music on video four. Do not regenerate immediately. Ask one question: is the problem the mood, the tempo, or the energy?

  • Mood wrong → new generation, different mood.
  • Tempo wrong → new genre within the same mood.
  • Energy wrong → keep the track, re-block the timeline.

That question saves whole regeneration cycles and makes the client’s feedback more useful next time.

What does this cost an agency?

Minutes are the currency. Paid plans start in the low tens of dollars a month for a set minute allowance, and a busy month of client work — say eight videos averaging three minutes plus two rounds of alternates — runs to roughly 50 generated minutes.

Above three or four clients, the pro tier pays for itself versus stock libraries where each track is licensed per project.

Where it does not fit

Broadcast and cinema delivery. The licences are built for online distribution; for TV or theatrical, a sync licence from a music publisher is the correct route.

Brands with an audio identity. If the client has a sonic logo or an exclusive composer, generated beds will not match.

Anything needing stems for a sound engineer to mix. You get a mixed track, not separated instruments.

Quick Summary

  • Keep a licence record per delivery: project, date, tier at generation, terms screenshot.
  • Build a per-client mood map (mood, intensity ceiling, instruments to avoid, approved references) to cut repeat work.
  • Duck music 15–20 dB under dialogue rather than asking for a quieter generation.
  • Diagnose revision requests as mood, tempo or energy before regenerating — re-blocking is free.
  • Paid plans start in the low tens of dollars monthly; online distribution only, not broadcast or stems work.