The DJ Instincts of a Real Radio Host, Automated
How It Works
A real radio station's programming discipline, running as an on-device scheduling engine.
Dual-Deck Equal-Power Crossfading
Two independent decks blend on an equal-power curve — automatically as a track nears its end, or dragged by hand on the crossfade bar — the same smooth blend a real board operator rides between two records.
On-Device Vocal & BPM Detection
Lightweight signal analysis finds exactly where each track's vocals begin and end, and estimates its tempo — entirely on the phone, no server round-trip required.
The 100-Track Plan
Every station builds a full 100-track queue up front against real programming rules: a target vocal/instrumental ratio, a rising-then-dipping BPM cycle every five tracks, repeat and artist cooldowns, and a guaranteed share of music the listener hasn't heard before.
DJ-Timed Track Intros
A spoken intro is timed to finish in the exact instant a track's own detected vocals begin — or, for a purely instrumental track, as close to 0:00 as possible — the same judgment call an FM host makes live, now computed from the track's own vocal onset.
Live Weather & News, Spoken
Apple WeatherKit forecasts and real news headlines are woven into that same intro, on whichever toggles the listener turns on — fitted to whatever time is actually available before the vocals start.
Aggregated From Jamendo & Audiomack
Stations mix results from Jamendo and Audiomack's independent/royalty-free catalogs into one continuous stream, with thumbs up/down and favorites remembered locally on the device.
Built From Real Airtime
FMify's most distinctive feature isn't a technical trick — it's software trained on four and a half years of actually doing the job.
Long before any of this was code, William Baldwin spent April 2007 to October 2011 on the air as a Radio Personality at FM KATCH Inc., and concurrently, from April 2008 to September 2010, at Aichi International Broadcasting Co., Ltd. — both in Nagoya, Aichi, Japan — hosting live shows and writing his own scripts.
That's where the instinct behind FMify's narration engine actually comes from. A real host doesn't just read a fact sheet over the intro — they time it: weather, a headline, the title and artist, landing and finishing in the exact instant the vocals come in, every single time, without a script reading long or the music getting stepped on. FMify's narration scheduler is that judgment call turned into an algorithm: it measures how long a track's own detected vocal onset actually gives it, fits as much of whatever's enabled (and due) into that window as it can, and always finishes on the title and artist right as the vocals begin — the same discipline from four and a half years of live radio, now running as Swift instead of a headphone cue.