Where Does the Speech in DJ Phonetic Come From?

DJ Phonetic draws its speech from historical recordings, and the page itself cites the Library of Congress as the source. You don't upload your own audio or record your voice: you pick from a set of preloaded speakers, and the site pairs each recording with a transcript and a phonetic breakdown.

What the page shows

The interface is built around a single flow: choose a speaker, then work with that speaker's audio. The visible elements on the page are:

• Select a speaker — a dropdown that loads the available voices ("Loading..." appears while it populates).
• Previous transcript / Next transcript — you move through the transcripts tied to the selected speaker rather than browsing a file list.
• Source: Library of Congress — the stated provenance of the recordings.
• Play audio / Stop audio — playback controls for the current clip.
• Toggle International Phonetic Alphabet — switches the phonetic view on or off.
• Text / Phones — the transcript text and its phone-level (speech-sound) representation.

So the answer to "where does it come from" has two layers: the recordings are historical speech held by the Library of Congress, and the site's own layer is the transcript plus IPA/phone data derived from that speech.

Why the source matters for the tool

DJ Phonetic's premise is that kicks, snares, and hi-hats are "hidden within human speech." That only works if the speech is clean, consistently transcribed, and phonetically annotated, because the drum sounds are mapped from the phones in the transcript. Historical archival audio from a national library fits that requirement better than arbitrary clips: the transcripts are already attached, and the phonetic layer can be generated against known text.

A practical example: if you select a speaker and step through transcripts with Next transcript, you're not just hearing a different voice — you're getting a different set of phones, which is what changes the beat you can build. The IPA toggle lets you see the phones that the drum mapping is working from.

What this means if you want to use it

• You cannot supply your own speech. The speaker list is fixed by what the site has loaded, so your creative range is the range of included speakers and transcripts.
• Expect archival audio, not studio vocals. The material is historical speech from the Library of Congress, so timbre and recording quality vary by speaker.
• Use the transcript controls to explore. Moving between transcripts is the fastest way to find a passage whose phones produce a beat you like.
• Check the IPA view when a beat sounds off. Seeing the phones explains why a given passage yields the hits it does.

The page does not state how many speakers are included or which specific recordings are used beyond the Library of Congress attribution, so treat the speaker dropdown as the authoritative list.