Check out our new blog post on music transformer that generates minute-long pieces that I find myself listening to again! It also generates compelling harmonizations and realizes chord progressions.
Which is your favorite? Ours is the Jazz 🎷
👂
I’m excited I’ll be joining MIT next fall, for a shared interdisciplinary faculty position between Music (
@MIT_MTA
@MIT_SHASS
) & EECS.
I’m recruiting PhD students
@MITEECS
for Fall 2024 (apply by Dec 15), and also Postdocs. See for details. Come join us!
Very honored to be a Canada CIFAR AI Chair, and also an adjunct prof
@UMontreal
!
I am recruiting postdocs+students+interns through
@Mila_Quebec
.
If you’re passionate about creativity+music+art, HCI, and ML, see and be sure to apply by Feb 1st, 2022!
Composer turned ML researcher Cheng-Zhi Anna Huang
@huangcza
(
@Mila_Quebec
,
@GoogleAI
) takes a human-centered approach to ML & creativity, expanding artists' creative reach. Huang also helped create Google’s 1st AI Doodle, the “Bach Doodle”.
Passionate about creativity+music+art, HCI, and ML?
I’m still recruiting PhD and master students through
@Mila_Quebec
for Fall 2023.
See for more details and apply by Dec 1st!
Remember the
#aisongcontest
? Follow our
#ismir2020
paper backstage to learn about the challenges musicians face when songwriting with AI, and how we might design ML to better support musicians’ needs and workflows.
Blog post:
Come be my colleague! Two faculty openings
@MIT_MTA
shared
Assistant prof w/
@MITMechE
@MITEECS
(Jan 15)
Tenured Associate/Full prof w/
@MIT_SCC
(Jan 2)
We’re starting a new graduate program in music tech
@MIT
! Come join us!
I’m excited I’ll be joining MIT next fall, for a shared interdisciplinary faculty position between Music (
@MIT_MTA
@MIT_SHASS
) & EECS.
I’m recruiting PhD students
@MITEECS
for Fall 2024 (apply by Dec 15), and also Postdocs. See for details. Come join us!
Only one more week to vote at the
#aisongcontest
!
38 songs, 38 teams of musicians + ML researchers, 38 different approaches to ML as a creative tool. Which one is your jam?
Cast your vote by EOD Wednesday June 30th!
▶️
The
#aisongcontest
is back! Register by 15 March. Submit by 1 June.
This year, we’ll have mentors to help teams get started!
🌱 To GET a mentor:
💜 To BE a mentor:
Details:
Be part of the AI Song Contest 2022! ➡️You can register your team until 15 March 🦾. If you are looking for team members - send us a message!
(In the picture you can see all the wonderful teams and cover art that brought the contest to life last year💜)
#aisongcontest
Only a few more days to vote at the
#aisongcontest
!
Check out this year’s entries to see how musicians + ml researchers are approaching human-ai co-creativity. Which one is your jam?
Cast your vote by Thursday June 30th!
▶️
Fun fact: the weights of the TensorFlow.js model that powers this doodle are only ~400KB
The whole doodle JS module, including TensorFlow.js, is ~200KB on the homepage (lots of room for improvement here in the future).
3/ We sped up Coconet from 40s to 2s in
#TensorFlowJS
by using dilated depth-wise separable convolutions, which requires less layers and are more accelerated then conventional convolutions. Also by fusing ops that are used in every Gibbs steps (
@GreenBeanDou
).
2/ The model Coconet 🥥 is an instance of OrderlessNADE and uses Gibbs sampling to generate the harmonizations through rewriting.
📜 Previous blogpost:
📝 Paper from
#ISMIR
2017:
tl; dr
1/ In three days, people spent 350 years worth of time playing with the Bach Doodle, and the “harmonize” button was clicked more than 55 million times.
6/ As the model was trained on Bach Chorales, melodies outside of the soprano range is “out of distribution” and harder to harmonize.
See how Paul Davids discovers this at 6:22 via
@YouTube
5/ We calibrated a speed test based on partial model evaluation time to determine if the harmonization request should be performed locally using
#TensorFlowJS
or sent to remote
#TPU
servers.
@jaxcooo
was able to get a nearly 50/50 split!
There were *21.6M* harmonizations submitted by people to the Bach Doodle!🙀
We are open sourcing this dataset along with a set of fun interactive visualizations to explore the data, discover some familiar top hits and even regional childhood songs! 🎵
📝
Mila is proud to welcome five new Canada CIFAR AI Chairs among its faculty. These world-class researchers will receive long-term funding to support their research activities.
7/ We found that parallel fifths and octaves were more common when user input was out of distribution, and fewer parallel fifths and octaves were correlated with positive user feedback.
The songs of the
#aisongcontest
all in one spot? 🤯 We made a playlist for you with many tracks of the contest on Soundcloud:
Listen and don't forget to vote for your favourite!