BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//programme.europython.eu//europython-2026//talk//7SSS93
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-europython-2026-7SSS93@programme.europython.eu
DTSTART;TZID=CET:20260717T110500
DTEND;TZID=CET:20260717T113500
DESCRIPTION:Music generation has gone from a research curiosity to somethin
 g you can try in a browser. Commercial platforms and open source models ca
 n produce full songs from a text prompt. Between the hype and the technica
 l papers\, it’s hard to get a straight answer about what’s actually go
 ing on  under the hood. **This talk is a clear\, honest walkthrough of how
  music generation systems work\, in simple language**\, no deep machine le
 arning knowledge needed.\n\nWe start with the core challenge: how do you t
 urn a continuous audio signal into something a generative model can work w
 ith? Neural audio codecs solve this by compressing waveforms into sequence
 s of discrete tokens\, and this idea is the foundation everything else bui
 lds on. From there\, we look at the two main modeling strategies: token pr
 ediction and diffusion. We compare what each does well\, where it struggle
 s\, and why the choice between them matters.\n\nOn the practical side\, we
  walk through the open source models and Python tools available today\, an
 d what you can build with them. Then we  get into evaluation\, one of the 
 most important open problems in the field. Current metrics only tell part 
 of the story\, and there is no standard benchmark for comparing systems. T
 his has real consequences for how research moves forward and how models ge
 t used.\n\nWe close with a discussion that often gets skipped: how artists
  and musicians see these tools\, what legal questions remain around traini
 ng data and copyright\, and why these conversations matter for the future 
 of the field.​​​​​​​​​​​​​​​​
DTSTAMP:20260524T130555Z
LOCATION:Theatre Hall (S2)
SUMMARY:How Music Generation Actually Works - Mateusz Modrzejewski
URL:https://programme.europython.eu/europython-2026/talk/7SSS93/
END:VEVENT
END:VCALENDAR