Upload any track. Our AI pipeline returns a Virality Score with a transparent breakdown, the exact seconds people will clip, your full Song DNA — plus 7 ready-to-post reels, captions, hashtags and a 2-week posting plan. Then paste your own TikTok or Reel to reverse-engineer what's working.
A single number 0–100 that rates your track's viral potential across TikTok, Reels, Shorts and Spotify — built from the song's own DNA: hooks, energy, structure and lyrics.
The exact 3–15 second segments of your song most likely to stop the scroll. Get timestamps, confidence scores, and drop them straight into TikTok or Reels edits.
Every measurable dimension of your track, extracted and benchmarked. The kind of data sheet labels use internally — now available to every artist.
Paste the link to your own TikTok, Reel or Short. Songbrain analyzes the visuals, the engagement curve and the comments — then cross-references everything with your song's hook map and tells you exactly what to change.
Every analysis renders 7 beat-synced 9:16 reels cut to your best moment — word-accurate karaoke-style lyric sync when the song has vocals, your cover art integrated, genre-matched color palettes. They render while the analysis runs: the first 3 are ready the second your results open.
AI transcription plus scoring on rhyme density, emotional impact, structural clarity and originality. See where your writing shines and where it falls flat.
Every instrument in the mix, identified — including world instruments like shamisen and guzheng. Production decisions, not guesswork.
Multi-model classification across fine-grained subgenres — so your reels, captions and hashtags match the scene your track actually belongs to.
Each stage runs a specialized model. Results combine into the final report.
Analytics tools only see what already happened. Playlist pitching is pay-per-submission. Songbrain is built on the opposite premise: score the song first, then hand you the content to promote it — 7 reels and a posting plan from a single upload.
Whether you're releasing your first EP or running A&R for a roster — the data speaks the same language. The interface adapts to how you use it.
You write songs, record in your bedroom, and release on DistroKid. Songbrain tells you — before release — whether your track has viral DNA, then hands you 7 ready-to-post reels and a 2-week posting plan so the promo is done before you finish your coffee.
You make 808s slap or shoegaze walls shimmer and post on Soundcloud. Songbrain breaks down tempo, mix translation, hook placement and production density straight from your track's DNA. Get the score + the reasons + what to fix.
You sign acts, pick singles, prioritize roster. Run every demo through Songbrain and you get objective, audio-driven inputs — Virality Score, hook placement, genre match, production readiness. A&R that scales.
The honest breakdown vs the alternatives indie artists and developers usually try first.
Spotify Audio Features API was deprecated in November 2024. ACRCloud is built for audio fingerprinting and recognition, not analysis. Essentia is an open-source library you need to self-host and chain manually — no virality prediction, no hosted API. Pay-to-pitch platforms like SubmitHub or Groover sell submissions; Songbrain turns the song itself into ready-to-post content instead.
The technical questions evaluators send before signing up — answered concretely.
The score is built from your song's own DNA — hook strength and placement, energy profile, structure, and lyric impact. Rather than predicting absolute virality — which no model can — it tells you how strong the track's shareable ingredients are, and the transparent breakdown shows exactly which components drive the number. Scores above 75 indicate strong viral ingredients; scores above 90 are exceptional. The detailed methodology is in our Virality Score explainer guide.
Each Best Moment window comes with a confidence between 0 and 1. The number combines four signals: (1) self-similarity matrix density — how often the section's melodic/rhythmic pattern repeats internally, (2) section-novelty edge strength — how distinct it is from surrounding context, (3) spectral centroid shift magnitude — how salient the audio brightness change is at that point, (4) energy contour position — where the moment sits on the rising edge of the track's energy curve. 0.85+ is a high-conviction hook; 0.65–0.85 is a candidate worth testing; under 0.65 is weak.
You paste an Instagram Reel or TikTok URL. Songbrain fetches the public engagement signals (per-second pause/replay clusters, comment timestamps, like-velocity spikes) and cross-references them with the audio at each timestamp. The output is a per-second breakdown: which moment landed, where attention dropped, what the audio-visual sync did at the peak. The analysis is free for videos under 7 days old (data is freshest then) and 30 credits for older ones.
MP3, WAV and FLAC up to 50MB per file. Most full-length songs sit well under 50MB at typical bitrates (320 kbps MP3 ≈ 8MB per 3 minutes; 44.1 kHz WAV ≈ 30MB per 3 minutes). For premium users, album uploads (up to 12 tracks per submission) are supported.
Whisper large-v3 runs the transcription, which supports 90+ languages with strong accuracy on the major ones (English, Spanish, German, French, Portuguese, Japanese, Korean, Hindi, Arabic). For very short snippets or heavily processed vocals, accuracy drops; the emotional scoring layer compensates by using the audio features when lyric confidence is low. Language is auto-detected — you don't need to specify it.
Yes. The REST API accepts one upload per call, but the architecture is built for parallel async processing — typical Premium quotas let you fire many concurrent uploads. Each job returns a `job_id` immediately; you can either poll `GET /api/v1/jobs/<id>` or supply a webhook URL for completion callbacks. Distributors and AI music platforms typically run hundreds of analyses per day this way. See the API docs for rate-limit specifics per plan.
Ours. The 10-model pipeline runs on dedicated cloud workers behind a queue-based job system. You upload the audio file, our infrastructure does the heavy compute (transcription, audio-feature extraction, genre classification, fusion), and you get the full report back. You never need to install anything or manage models — and your reels render on our side while the analysis runs.
Upload your first track in under a minute. Get your Virality Score, Best Moments, full Song DNA — plus 7 ready-to-post reels and a posting plan.
Analyze your song free →