Kyle Robinson @Artt__Attack

beep boop bleep blop HBSc @myLakehead ➡️ MMath @UWCheritonCS ➡️ ML @ VoyagerX Seoul, Korea Joined December 2011

Tweets

3K
Followers

129
Following

180
Likes

4K

Kyle Robinson @Artt__Attack

4 years ago

@keunwoochoi I say we take an FFT of the spectrogram images and train on that.

0 0 3 227 0

View Details

Kyle Robinson @Artt__Attack

4 years ago

@urinieto @NicholasJBryan @j_p_caceres Any plans to add this to pip or conda directly?

1 0 1 0 0

View Details

Kyle Robinson @Artt__Attack

4 years ago

@urinieto @NicholasJBryan @j_p_caceres This was so perfectly timed 😍. Just this week I was looking for a fast torch/ONNX mel layer for a TTS cloning deployment. <3 thanks all

1 0 1 0 0

View Details

Kyle Robinson @Artt__Attack

4 years ago

First @ISMIRConf missed since my first in 2019 🥹. Have fun in India everyone! Miss you all~

0 0 1 0 0

View Details

@JaguarWorldwide @SonyMusicUK Great to see a report like this :). Shared by a friend. I have one question tho. The photo above says 99% 'male primary artist', but the report says 99% 'include male artist', 76% 'male primary artist'. Still a shocking statistic ofc 😱. Can you clear this up?

0 0 1 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@birnsi omggggggg

0 0 3 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@birnsi Wot

0 0 0 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@birnsi Who

1 0 0 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@ismir2021 making me want to get back into MIR/MusicRec🥲. I love this community so much. Hope to see everyone in person next year!

0 1 7 0 1

View Details

Kyle Robinson @Artt__Attack

5 years ago

@m_schedl @u_bode @cpjku @CS_JKU @jkulinz Congrats! Really looking forward to seeing the work that comes from this :). If I was at another point in my career I'd be fighting for a spot!

0 0 2 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@birnsi 🥺

0 0 1 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@psobot Ahahaha lots of Korean sites make you install some old security software before you proceed, but you can usually just change the css to make the button visible.

0 0 0 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

If @theweeknd's production is now future 80's, and he's popular, and we are technically living in the future, are we finally IN THE FUTURE 80's?! I hope cyberpunk is next.

0 0 2 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@birnsi @tywooping @Yvonner8r 😒

0 0 2 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@CanRevAgency logic: -Let's enforce 2fa! 👍 -Let's only use phone numbers 😐 -Let's ask if they have a foreign phone number 👍 -If they do let's force them to call in during Canadian business hours to unlock their account 😑

0 0 0 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@benhayesmusic Super dope! Wish we'd have the opportunity to chat in person! Great work :)

0 0 0 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

The final paper from my masters work at @Uwaterloo on music recommendation diversity accepted to @ismir2021 :). Wish it was in-person, but looking forward to ismir #3!

1 2 31 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

The craziest experience in surrounding yourself with another language is how it affects your first language. Today I said 'I hope you enjoy it well~' which when I think about it is kinda weird in English?

0 0 7 0 0

View Details

Kyle Robinson @Artt__Attack

5 years ago

@ervindervishaj I haven't unfortunately. Ultimately they didn't work great for the kind of clustering I was trying to do so I moved on. Maybe I'll get around to it sometime, but I doubt anytime soon 🥲

0 0 0 0 0

View Details

Kyle Robinson @Artt__Attack

6 years ago

Visualizations of the first 3 principal components of latent track features from matrix factorization (ALS) on about 20k songs from my listening history. Color is the popularity of the track in the matrix. First is the raw features and the second is L2 normalized by sample.