Last month a client asked me to deliver a 90-second product demo by Friday. The footage was cut, the voiceover was recorded, and the only missing piece was music. I opened Soundraw at 2:14pm. By 2:32pm I had a finished track sitting under the edit: timed to the second, cleared for commercial use, with separate stems I could duck under the narration.
That turnaround isn’t luck. Soundraw is an AI music generator that works best when you treat it less like a slot machine and more like a session musician who needs a clear brief. Here’s the workflow I use now, in order, including the actual settings and numbers.
Step 1: Write the brief before you touch the prompt bar
The biggest time-waster with any AI music tool is generating forty tracks and hoping one sticks. Spend three minutes writing down five things:
- Exact duration: 90 seconds, 22 seconds, 6 minutes. Not “about a minute.”
- Tempo feel: slow and spacious (70–90 BPM) or driving (110–125 BPM).
- Emotional arc: does it need to rise, or sit still and stay out of the way?
- Where the voice sits: with narration you need space in the 200 Hz–4 kHz range.
- Moment to hit: the logo reveal at 0:12, the product shot at 1:04.
For my demo video the brief read: 90 seconds, 105 BPM, uplifting but not triumphant, no vocals, peak energy from 0:30 to 1:10. One line of context narrowed the search before I’d generated anything.
Step 2: Pick genre and mood for the edit, not your taste
Soundraw organises generation around genre and mood tags plus a length slider. The temptation is to browse until you find something you personally like. Don’t. Pick for the picture.
A software demo usually wants Cinematic > Inspiring or Electronic > Uplifting. A testimonial video wants something warmer and quieter, like Acoustic > Hopeful. If the visuals are fast-cut and punchy, go higher tempo and let percussion lead. If you’re scoring a talking-head interview, avoid anything with prominent synth stabs, because those fight the dialogue on every cut.
A quick test for “will this sit under speech?”
Play the candidate track at 30% volume while reading the script aloud. If you catch yourself unconsciously raising your voice to be heard, the track is too busy. Move on to the next generation.
Step 3: Use the section editor like a video timeline
This is where Soundraw separates itself from one-click generators. After generating, you get an arrangement view where the track is broken into sections, typically intro, verse, chorus, bridge and outro. You can swap, duplicate, delete or reorder them. If you’ve ever edited audio, it looks familiar. If you haven’t, think of it as Lego bricks of music.
Two moves pay off immediately:
- Trim the intro to two bars when your video opens on a hard cut. Long ambient intros feel amateurish on social.
- Duplicate the chorus to extend the emotional peak instead of letting the track drift into a bridge right when your product is on screen.
If you want a closer look at how much manual control you actually get, and where it still feels automated, the full Soundraw review walks through the interface section by section.
Step 4: Lock the length to the frame
Set the duration slider to your exact runtime before you download. A 90-second video needs a 90-second track, not a two-minute one you fade out at the end. The fade always lands in the wrong place.
Do the bar math so your edits land where you want them. At 105 BPM, one 4/4 bar lasts about 2.29 seconds, so 90 seconds is roughly 39 bars. If your product reveal happens at 0:12, that’s bar 6. A two-bar intro plus a four-bar build drops the chorus almost exactly on the cut. Small numbers, big difference.
Step 5: Download stems, not just the master
Soundraw exports both a full mix and individual stems, depending on your plan. Always grab the stems. On a 90-second demo with continuous voiceover, I typically:
- Drop the melody stem 3 dB during narration and bring it back in the gaps.
- High-pass the full mix at 180–200 Hz so the bass doesn’t muddy the voice.
- Keep the drum stem at full level through the product shots for momentum.
That’s a five-minute mix job, and it’s the difference between a track that sounds intentional and one that sounds stock.
Worked example: the 90-second product demo
Here’s how the timeline actually broke down.
- 0:00–0:06 — Two-bar intro, drums only, under a wide shot of the office. Low energy on purpose.
- 0:06–0:30 — Verse section. Bass and synth pad enter while the voiceover explains the problem the product solves.
- 0:30–1:10 — Chorus, full arrangement. This is the feature walkthrough, so the music peaks and the melody stem is ducked 3 dB under the voice.
- 1:10–1:22 — Bridge. The arrangement thins out and one final line of narration lands cleanly.
- 1:22–1:30 — Outro held on a single chord for the logo and call to action.
Shorter edits compress the same way. There’s a useful breakdown of scoring a 90-second product demo with Beatoven.ai that shows a different tool tackling the identical brief, and the section-by-section thinking carries straight over.
Mistakes that make AI music sound like AI music
- Using the first generation. Make at least five variations of the same brief. The gap between them is wider than you’d expect.
- Running the track at full volume under dialogue. Duck it, every time.
- Ignoring tempo. Match the BPM to your cut rhythm and the whole edit feels tighter, even when nothing else changed.
- Fading out mid-phrase. Use the section editor to end on a resolved chord instead.
Where Soundraw fits in a bigger AI music stack
Soundraw’s strength is structure and control: you decide the arrangement, the length and the section order. Other tools lean harder into instant results. If speed matters more than arrangement, a six-step Boomy workflow for turning a raw AI idea into a finished track gets you there faster. And if most of your work is narration-heavy, this Beatoven.ai review covers where it beats Soundraw and where it doesn’t.
None of these replace a composer on a paid brand film. For YouTube videos, course modules, podcast intros and client explainers on a deadline, they’re more than good enough.
Template your settings so the next track takes eight minutes
The trick to getting fast with Soundraw isn’t memorising the interface. It’s reusing what already worked. After that first demo video I saved the brief — 105 BPM, Cinematic > Inspiring, 90 seconds, four-section arrangement — as a note in the project folder.
The next video in the same series took eight minutes from open to export. Same tempo, same mood, new melody, and it felt like a deliberate brand sound rather than a random track pulled from a library. If you publish regularly, build three or four of these templates: one for high-energy promos, one for calm tutorials, one for product launches. Then you’re not generating music from scratch every time. You’re picking a lane and hitting go.