# TrackSensei

## About
TrackSensei is an AI production mentor for electronic music producers. Upload an Ableton Live project (.als) or an audio bounce and get genre-aware feedback on arrangement, sound design, mix, and master — plus a chat mentor that has actually read the analysis of your specific track. Built for producers who want concrete, numerical feedback instead of vague "sounds good, bro" reviews.

## What TrackSensei Does
- Parses Ableton Live `.als` project files directly in the browser to inspect tracks, groups, devices, automation, and arrangement structure.
- Analyzes audio bounces (WAV, MP3, AIFF, FLAC, M4A) for tonal balance, LUFS loudness, true peak, stereo width, mono compatibility, transient behavior, and spectral distribution.
- Runs Stem Interaction Analysis on multi-stem uploads to evaluate how kick, bass, mids, and tops actually fit together.
- Produces a Deep Dive: a written, genre-calibrated review that explains what is working, what is off, and exactly which numbers are out of range for the target style.
- Hosts a chat mentor that already has the full analysis context loaded — you can ask "why is my low end muddy" and it will answer against your specific track's numbers, not generic advice.
- Scores tracks against calibrated genre profiles covering roughly 80 subgenres across 12 main families.

## Plans and Pricing
- **Free — €0/month.** 1 project per month, up to 3 track versions per month with AI feedback on every version. No credit card required.
- **Pro — €12/month.** 150,000 feedback tokens per month (roughly 12 full track journeys, or 40+ quick analyses). Marked Most Popular on the pricing page.
- **Ultimate — €29/month.** 400,000 feedback tokens per month (roughly 32 full track journeys, or 110+ quick analyses), Stem Interaction Analysis included.
- **Boosters** (one-time, never expire): roughly €15 (S), €30 (M), €60 (L) for additional feedback tokens, available to Ultimate subscribers and consumed only after the monthly budget runs out.
- Every AI request (a Deep Dive is about 3,500 tokens, a chat message about 1,500) draws from the same monthly token budget; a typical track journey of 1 Deep Dive plus 6 chats is about 12,500 tokens.
- All prices are in euros, billed via Stripe. EU VAT is added where applicable.

## Genres Supported
TrackSensei ships 12 main genre families with around 80 subgenres total. The six calibrated electronic profiles below have explicit per-genre LUFS, sub (20–60 Hz), bass (60–250 Hz), mids, and mono-compatibility targets derived from a 563-track reference corpus.

- **Techno.** Peak-time techno, Berlin groove, melodic techno, hard techno, raw techno, hypnotic techno, industrial, schranz.
- **House.** Deep house, tech house, soulful house, jackin' house, garage house, funky house, disco house.
- **Trance.** Uplifting, progressive, psy-trance, tech-trance, hard trance.
- **Drum & Bass.** Liquid, neurofunk, jump-up, deep DnB, halftime.
- **Dubstep.** Brostep, riddim, melodic dubstep, deep dubstep, 140.
- **Hardstyle.** Euphoric, raw, rawphoric, uptempo.
- **Hardcore.** Frenchcore, terror, gabber, UK hardcore.
- **Hip-Hop / Trap.** Boom-bap, modern trap, drill, lo-fi hip-hop.
- **Breaks.** UK breaks, breakbeat, nu-skool breaks.
- **Ambient.** Dark ambient, drone, downtempo, lo-fi.
- **Pop.** Modern pop, electro-pop, synth-pop, indie-pop.
- **Rock.** Indie rock, alt-rock, post-rock.

The six profiles with corpus-calibrated numeric targets (May 2026 calibration) are: house, tech-house, techno-berlin-groove (peak-time), techno-melodic, techno-hard, and techno-raw.

## How It Works
1. Upload a `.als` project file or an audio bounce. Audio decoding and feature extraction happen entirely in your browser — the raw audio bytes never reach a server.
2. The client computes loudness (LUFS integrated and short-term), true peak, sub/bass/mid/high band energy, stereo width, mono LUFS delta, transient stats, and structural features.
3. The extracted features JSON is sent to TrackSensei's backend, which calls Claude with the matching genre profile to generate the Deep Dive review and score.
4. You receive a per-section breakdown (arrangement, sound design, mix, master), a numeric score against the genre target, and a written critique with specific fixes.
5. You can then chat with the mentor about that exact track — it keeps the analysis in context so follow-up questions get specific answers, not generic advice.

## What Makes TrackSensei Different
Audio and `.als` parsing are 100% client-side — your tracks never leave your machine, which matters if you care about unreleased work staying unreleased. The genre profiles are calibrated against a 563-track reference corpus (54 house, 38 tech-house, 57 berlin-groove, 56 melodic, 106 hard, 202 raw, 50 misc), so the LUFS windows and band-energy targets reflect what actually charts on Beatport in each style rather than generic loudness-war thresholds. The chat mentor has the full feature JSON of your specific upload, which means "your kick is at -8.2 LUFS short-term, that is 2 dB hot for raw techno" instead of "try turning the kick down a bit."

## Frequently Asked Questions

**What is TrackSensei?**
TrackSensei is an AI production mentor for electronic music producers. You upload an Ableton project or an audio bounce, and it returns a genre-aware analysis covering arrangement, sound design, mix, and master, plus a chat mentor that has read the full analysis of your specific track.

**How much does TrackSensei cost?**
Free is €0/month with 1 project and up to 3 track versions per month. Pro is €12/month with 150,000 feedback tokens per month (roughly 12 full track journeys). Ultimate is €29/month with 400,000 feedback tokens per month (roughly 32 full track journeys) and Stem Interaction Analysis included. Boosters of roughly €15, €30, and €60 add extra tokens for Ultimate subscribers and never expire.

**Does TrackSensei work with Ableton Live projects?**
Yes. You can upload an `.als` file directly. TrackSensei parses the project in the browser to inspect tracks, groups, devices, automation, and arrangement structure, then includes that structural data in the Deep Dive review. No audio is rendered server-side.

**What audio formats does TrackSensei support?**
WAV, MP3, AIFF, FLAC, and M4A. Stereo and mono are both supported. For Stem Interaction Analysis (Ultimate plan), upload separate stems for kick, bass, mids, and tops.

**Does my audio get uploaded to a server?**
No. Audio decoding, `.als` parsing, and feature extraction all run in your browser via the Web Audio API. Only the resulting numeric feature JSON is sent to the server so Claude can generate the written review. The raw audio bytes never leave your device.

**What genres does TrackSensei analyze?**
Twelve main families covering around 80 subgenres: techno, house, trance, drum & bass, dubstep, hardstyle, hardcore, hip-hop/trap, breaks, ambient, pop, and rock. Six electronic profiles — house, tech-house, techno-berlin-groove, techno-melodic, techno-hard, and techno-raw — have corpus-calibrated numeric targets for LUFS, sub, bass, mids, and mono compatibility.

**How accurate is the genre-aware scoring?**
The six calibrated profiles use empirical P25–P75 windows derived from a 563-track reference corpus of released material in each genre. That means the LUFS and band-energy targets reflect what actually gets released and played out, not theoretical mastering rules. Less common genres fall back to broader defaults, so scoring is most precise for the six calibrated electronic profiles.

**Can I get feedback on my arrangement, not just mixing?**
Yes. The Deep Dive covers four sections: arrangement, sound design, mix, and master. Arrangement feedback uses both audio-derived structural cues (energy curve, section boundaries, transient density) and, if you upload an `.als`, the actual scene and clip layout from the project file.

**Is there a free plan?**
Yes. The Free plan is €0/month with no credit card required, and includes 1 project and up to 3 track versions per month with AI feedback on every version. It uses the same analysis pipeline as paid plans — the only differences are the monthly limits and access to Stem Interaction Analysis (Ultimate only).

**What's the difference between a Deep Dive and a chat message?**
A Deep Dive is the full upload-and-review cycle: you submit a track, the analyzer runs, and you get a written multi-section review with a score. A chat message is one back-and-forth with the AI mentor after the Deep Dive is done. Both draw from the same monthly pot of feedback tokens (a Deep Dive is about 3,500 tokens, a chat message about 1,500), so one Deep Dive can be followed by many chats about that same track before the budget runs out.

## Contact
- Website: https://tracksensei.com
- Email: hello@tracksensei.com
- Twitter / X: @tracksensei

## Markdown Mirrors (Clean AI-Readable Versions)

Every page on this site has a plain-markdown mirror. Add .md to any page URL for navigation-free content:

- https://tracksensei.com/index.md
- https://tracksensei.com/pricing.md
- https://tracksensei.com/faq.md
- https://tracksensei.com/about.md
- https://tracksensei.com/contact.md
- https://tracksensei.com/register.md
- https://tracksensei.com/privacy-policy.md
- https://tracksensei.com/terms-of-service.md