Keunwoo Choi @keunwoochoi Twitter profile

Pinned Tweet

Keunwoo Choi

5 months

hi music people, i wrote a tutorial on large language models and music information retrieval. of course it's called.. LLMs <3 MIR 🥁 have fun!

1

28

194

Last Seen Profiles

@lifetosmoochy

@clinefiles

@memekIbu2

@JeseniaWhyte

@NekotaNyankiti

@InspekSolutions

@mikaelaa011

@gar_mais

@bestofmargot

@couples_mot7arr

@AndreaHolowka

@heirenshuaiba

@iMsolla

@chumboy666

@stw_pdg

@BaselSharafMD

@VmaniakJ

@FukuzumiTakashi

@CamachoCirce

@memekIbu2

@yemen775050

@dngjng177373

@stwmaniax

@LylaTapaoa40081

@thegbenuola

@cole_albertt

@Safia_Alsnaani

@stw_pdg

@Navyrecife13

@FukuzumiTakashi

@siamzachon

@AriahK27796

@host

@OldOrderOfAges

@BCUsoftball

@memekIbu2

Keunwoo Choi

@keunwoochoi

2 years

whoa, this is bigger than ChatGPT to me. google almost solved music generation, i'd say.

150

1K

6K

Keunwoo Choi

@keunwoochoi

5 years

initialize yourself with it!

2

306

595

Keunwoo Choi

@keunwoochoi

2 years

One of the key models in MusicLM is SoundStream, an audio codec. It made vocoders obsolete; and reshaped audio generation as a token prediction task. SS is not open to public, but a similar neural audio codec Encodec is completely open-source →

GitHub - facebookresearch/encodec: State-of-the-art deep learning based audio codec supporting both...

State-of-the-art deep learning based audio codec supporting both mono 24 kHz audio and stereo 48 kHz audio. - facebookresearch/encodec

github.com

7

48

400

Keunwoo Choi

@keunwoochoi

7 years

I won the best paper award in #ismir2017 !!! Feeling honoured!!!! Thanks for co-authors @markbsandler György Fazekas @kchonyc

21

20

260

Keunwoo Choi

@keunwoochoi

2 years

really well done, from SoundStream and AudioLM through MuLan to MusicLM 👏👏 the overall structure of MusicLM = MuLan + AudioLM = MuLan + w2v-BERT + SoundStream

2

21

251

Keunwoo Choi

@keunwoochoi

5 years

Animated attention by @kchonyc which took his good 3hrs @facebookai @PyTorch #torched #deeplearning #nlp #attention #seq2seq #kyunghyuncho #nyu #cds

3

11

194

Keunwoo Choi

@keunwoochoi

2 years

MuLan is a text-music joint embedding model. - contrastive training - 44M music audio - text description pairs from "internet music videos" *cough cough* youtube *cough cough* - AST: audio spectrogram transformer

5

11

134

Keunwoo Choi

@keunwoochoi

2 years

hi all, here's an academic proof that AI has peaked in 2021 and started to downturn by 1.346% in 2022. diff = np.log(np.exp((1 - 138490 / 140380)))

17

8

136

Keunwoo Choi

@keunwoochoi

2 years

DALL·E mini - "spectrogram of rock music"

13

9

127

Keunwoo Choi

@keunwoochoi

4 years

Last Friday was my last day of the two years at Spotify. I started to work at ByteDance AI Research from today. (At Mountain View (California) in principle, but joined remotely from NYC)

12

0

123

Keunwoo Choi

@keunwoochoi

3 years

I left ByteDance last Friday. It was such a 1.8 year ❤️ (base-12) I'm glad I got what I wanted - a novel and intense learning experience. I shipped quite a few stuff, worked on research back-end tools, and made some research impact. Now, time to move on :)

5

2

117

Keunwoo Choi

@keunwoochoi

10 months

🌱 We’re hiring 2024 summer research interns on LLMs for drug discovery and biomedical applications. Join me, @stephenrra , @kchonyc , and other amazing people at NYC to work on the LLM product development of @PrescientDesign , @genentech ✨ Details:

0

21

115

Keunwoo Choi

@keunwoochoi

5 years

nnAudio: #pytorch CQT layers + etc. Done by Kin Wai Cheuk et al. And yes, it’s fast.

3

24

105

Keunwoo Choi

@keunwoochoi

2 years

🥳 PROPOSAL: Foley Sound Synthesis Challenge 🥳 There are enough challenges out there for speech and music. We propose one for "the other" kind of audio -> sound. Or effects. Or, Foley. We need to define the problem, dataset, and eval scheme. How? 🧵🧶

A Proposal for Foley Sound Synthesis Challenge

"Foley" refers to sound effects that are added to multimedia during post-production to enhance its perceived acoustic properties, e.g., by simulating the sounds of footsteps, ambient environmental...

arxiv.org

9

19

105

Keunwoo Choi

@keunwoochoi

1 year

I summarized the difference between `tokenizers.Tokenizer`, `transformers.PreTrainedTokenizer`, and `transformers.PreTrainedTokenizerFast`. I even made a github repo just to post this.

GitHub - keunwoochoi/tokenizer-vs-tokenizer

Contribute to keunwoochoi/tokenizer-vs-tokenizer development by creating an account on GitHub.

github.com

1

18

101

Keunwoo Choi

@keunwoochoi

3 years

All you need is AI and music -- I'm giving a guest lecture today at NYU, Center for Data Science. Stay tuned for the recording and slides :)

6

99

Keunwoo Choi

@keunwoochoi

3 years

Ahem, ahem. : I joined Gaudio Lab to - i'd dare say - pioneer some audio/music AI! 🥳 I'm excited more than ever :D Oh, and I'll visit Seoul more often. Friends in 🇰🇷, catch up soon!

Gaudio Lab | Where Sound Is

Gaudio Lab delivers innovative audio solutions to platforms, device manufacturers, and content creators.

www.gaudiolab.com

5

2

98

Keunwoo Choi

@keunwoochoi

7 years

@OHnoTypeCo

2

6

95

Keunwoo Choi

@keunwoochoi

2 years

+ they released MusicCaps dataset (5521 music-text pair) which they used as an eval set. .

MusicCaps

5.5k high-quality music captions written by musicians

www.kaggle.com

6

7

92

Keunwoo Choi

@keunwoochoi

1 year

the “llama moment” has come to audio research today! i can’t even imagine what we’ll see out of AudioCraft. whatever you work on in music/audio, do consider using it, as much as you can. if you don’t know what to do, think what you can do with it and get a head start.

AI at Meta

@AIatMeta

1 year

Today we're sharing details on AudioCraft, a new family of generative AI models built for generating high-quality, realistic audio & music from text. AudioCraft is a single code base that works for music, sound, compression & generation — all in the same place. More details ⬇️

39

529

2K

4

11

92

Keunwoo Choi

@keunwoochoi

11 months

THIS IS BIG! All the music folks in Google Deepmind focus on one thing: AI music generation while NOT exploiting artists. Nothing is perfect, there're probably still some holes in giving the credit, but this is better than anything ever for very sure.

Demis Hassabis

@demishassabis

11 months

Thrilled to share #Lyria , the world's most sophisticated AI music generation system. From just a text prompt Lyria produces compelling music & vocals. Also: building new Music AI tools for artists to amplify creativity in partnership w/YT & music industry

111

528

3K

2

8

92

Keunwoo Choi

@keunwoochoi

3 years

"All you need is AI and music" by Keunwoo Choi, 2021-12-08; A guest lecture at New York University - at @kchonyc 's class. Now you can watch it :)

DS-GA 1011 (Fall 2021) - Lecture 13 - AI for Music and Music for AI...

Presenter: Keunwoo Choi (https://keunwoochoi.github.io)Slides: https://www.slideshare.net/KeunwooChoi/all-you-need-is-ai-and-music-by-keunwoo-choi---------- ...

www.youtube.com

4

17

91

Keunwoo Choi

@keunwoochoi

2 years

New AI music model alert! yes, again 🎉 #SingSong , another music generation model by Google; @chrisdonahuey et al. Ok let me do another run for collecting followers. How does it work?

1

11

90

Keunwoo Choi

@keunwoochoi

4 years

If you belong to an underrepresented group in any sense (gender, race, nationality, financial situation, etc) and need some help on any MIR issues, please just contact me. gnuchoi at the-email-starting-with-G-you-know-what-I-mean😉

4

12

88

Keunwoo Choi

@keunwoochoi

6 years

My #NeurIPS2018 summary from music x ML (=music information retrieval (=MIR)) perspective is here:

Machine Learning for Creativity and Design Workshop (NeurIPS2018), and +@

Following their first workshop last year, there was the second ML4 Creativity and Design workshop on 8th Dec 2018 at Neurips2018 (=one of the biggest machine learning conferences), Montreal (=one o…

keunwoochoi.wordpress.com

3

29

85

Keunwoo Choi

@keunwoochoi

7 years

Hi all, I'm happy to twit-announce that I'm joining 🎧 Spotify NYC from June! 😀

13

7

82

Keunwoo Choi

@keunwoochoi

7 months

for #icassp2024 attendees, i'm open sourcing my `What to eat around COEX` list. originally written for @cwu307 but sharing it for a large crowd and make the world better place, reduce p(doom), etc.

What to eat around COEX - Keunwoo's recommendation, 2024 Edition

What to eat around COEX Written by keunwoo or keunwoo.OOO🎶 Hi ICASSP attendants, Unfortunately I can’t make it to ICASSP 2024 but may my gourmet soul be with you! Please feel free to share this with...

docs.google.com

2

14

81

Keunwoo Choi

@keunwoochoi

3 years

📄+📄+📄+📄+📄+📄+📄= 7 papers 🔥MIR researchers at ByteDance (SAMI team) made 7 papers accepted to #ISMIR2021 🔥 🧵I'll introduce them here one by one :)👇

1

6

79

Keunwoo Choi

@keunwoochoi

5 years

Hi people! Me and @kchonyc 's #ismir2019 paper, "Deep Unsupervised Drum Transcription" aka 🥁 DrummerNet is here. Paper --> Blog post --> Supplementary material -->

GitHub - keunwoochoi/DrummerNet: Supplementary material of "Deep Unsupervised Drum Transcription",...

Supplementary material of "Deep Unsupervised Drum Transcription", ISMIR 2019 - keunwoochoi/DrummerNet

github.com

2

18

78

Keunwoo Choi

@keunwoochoi

2 years

NeurIPS review complete award continues. Sponsor: @kchonyc

1

78

Keunwoo Choi

@keunwoochoi

2 years

to recap, i find the whole roadmap really, really brilliant. - because there's MuLan, they could use audio-only dataset. - because there's SoundStream, the music generation task was simplified to token generation, not waveform generation.

3

75

Keunwoo Choi

@keunwoochoi

2 years

Ok now (restrospectively, on high-level) it's kinda simple. given an training item: - extract MuLan tokens (M), extract w2v-BERT (S), SS tokens (A) - train model for M → S. - train model for [M;S] → A both done by decoder-only transformers.

1

4

67

Keunwoo Choi

@keunwoochoi

9 months

i'm teaching a class about AI at NYU, Spring 2024. it's "Deep Learning for Media", a course about AI for audio and visual contents. oof, i thought i became an LLM person. (it's not a job change, i'm covering one class this semester) happy to find back a nyu dot edu account!

5

2

69

Keunwoo Choi

@keunwoochoi

3 years

GSEP - Gaudio Source Separation 🔥🔥🔥

6

28

66

Keunwoo Choi

@keunwoochoi

6 years

Audio/Deep learning folks - please go check out our `torchaudio-contrib`! By @faroit , Kiran Sanjeevan, and me.

GitHub - keunwoochoi/torchaudio-contrib: A test bed for updates and new features | pytorch/audio

A test bed for updates and new features | pytorch/audio - keunwoochoi/torchaudio-contrib

github.com

3

27

66

Keunwoo Choi

@keunwoochoi

1 year

👋 I joined @PrescientDesign recently. I distracted @kchonyc with music research circa 2016-2019. This time he offered me to join his realm -- languages! I'm already having a lot of fun, knowing more to come.

6

2

65

Keunwoo Choi

@keunwoochoi

3 years

@Gnarbone @AskAKorean

0

2

62

Keunwoo Choi

@keunwoochoi

3 years

MT3: Multi-Task Multitrack Music Transcription T5, but for music transcription. A neat solution to cope with many-but-small existing datasets.

1

6

62

Keunwoo Choi

@keunwoochoi

3 years

<shameless as always> my papers are 1st and 6th most cited ISMIR paper in the last 5 years!🔥🔥 heard it was mentioned at the #ismir2021 trivia organized by the titans @r4b1tt @urinieto . i think they should arXiv the trivia and cite my paper thx

5

2

61

Keunwoo Choi

@keunwoochoi

1 year

🎙 Let's talk about AI research. And datasets. Accessibilities. Opportunities. Music.

1

6

61

Keunwoo Choi

@keunwoochoi

2 years

ChatGPT blew up because people *hate* writing 😂

3

2

61

Keunwoo Choi

@keunwoochoi

2 years

AudioLM = w2v-BERT + SoundStream w2v-BERT is.. - a BERT, but for audio. originally for speech. in AudioLM, an intermediate layer from speech-pretrained model was used. - it's "coarse" (250bps of bitrate.) - it takes care of semantic information.

1

0

57

Keunwoo Choi

@keunwoochoi

4 years

ByteDance/TikTok is hiring research scientists and software developers around music information retrieval and music/audio signal processing at Mountain View, US. Please hit me up! #ismir2020

1

12

57

Keunwoo Choi

@keunwoochoi

2 years

SoundStream is.. - a neural audio codec. - residual vector quantizer (RVQ) is used - as a codec, it's "fine-grained" (2000bps of bitrate)

2

0

52

Keunwoo Choi

@keunwoochoi

8 months

we're hiring AI/LLM engineers! - covering both pre-training and post-training tasks - purely for product development, based on *extensive understanding in LLMs* - with real-world impacts on drug discovery in Genentech - no publication within sight

1

12

54

Keunwoo Choi

@keunwoochoi

5 years

@urinieto ROCKING #ismir2019 HAHAHAHAHAHA 😂😂😂 seriously, my every follower should watch this otherwise please unfollow thanks.

0

12

53

Keunwoo Choi

@keunwoochoi

3 years

my code was more interesting 4+ years ago.

3

0

52

Keunwoo Choi

@keunwoochoi

5 years

Frrquency-aware CNNs. Ooops I was working on the same thing last summer but had no time after some experiments. It worked for music classification and source separation. Go try this!

4

3

51

Keunwoo Choi

@keunwoochoi

2 years

do you know what ChatGPT can't do? 🔊 audio generation. we do, at Gaudio Lab 😉

2

10

53

Keunwoo Choi

@keunwoochoi

6 years

What would you say if I passed the PhD viva today? I mean, I did, so feel free to really say it!

21

0

51

Keunwoo Choi

@keunwoochoi

3 months

i like textgrad. down to trying it. but i can't really say i like the way it's explained.. paper/blogpost is purely an analogy to backpropagation, which is cool but can you also just simply describe what it is..?

5

2

52

Keunwoo Choi

@keunwoochoi

2 years

🎉 It's happening. Foley Sound Synthesis Challenge! Generative AI folks, join us and make some sound! 🔊

1

14

50

Keunwoo Choi

@keunwoochoi

3 years

We're looking for a junior-level MIR researcher (perhaps Master or PhD) in Shanghai; to work with me on music tagging and related problems. Expecting to hire ASAP. Please email me if you're interested!

2

12

48

Keunwoo Choi

@keunwoochoi

4 years

It seems clear to me that Tensorflow developers are not deeply understanding why researchers struggle with their product. Life is too short for most of researchers to be very good at all Python and machine learning. TF adds another burden, but Pytorch doesn't.

4

10

48

Keunwoo Choi

@keunwoochoi

2 years

ISMIR2022 tutorials are out! 👉

1

11

49

Keunwoo Choi

@keunwoochoi

7 years

your code vs my code 😎 #swag

2

10

47

Keunwoo Choi

@keunwoochoi

2 years

in the training set, no text label is needed because we.. i mean, googlers.. have pre-trained MuLan! also, if you believe the power neural codec, SoundStream, no need to trained end-to-end with waveforms etc! SoundStream tokens are good enough!

1

0

46

Keunwoo Choi

@keunwoochoi

7 years

U-net with 33 lines with #keras . How can I not love this API?

1

8

44

Keunwoo Choi

@keunwoochoi

2 months

it took me 4 years to get started learning about NYC / Brooklyn jazz scene. now i'm totally immersed in it, after attending ~100 shows in the last 2 years. subscribe my newsletter "JazzBuzz" and learn about the captivating world - people, music, venues!

1

4

48

Keunwoo Choi

@keunwoochoi

3 years

TikTok🎶 is hiring a research scientist in Music/ML @🇬🇧 London office 🔥 Join our SAMI team to work on Speech, Audio, and Music intelligence with us :) Please feel free to reach out to me for any question 📧

0

6

46

Keunwoo Choi

@keunwoochoi

3 years

*QUITE A FEW* papers are accepted to #ismir2021 from our team in ByteDance 🚀🚀🚀🚀🚀🚀🚀 I'll share more details once the proceedings are updated. And yes we're hiring 🔥🔥🔥🔥🔥🔥🔥

0

2

45

Keunwoo Choi

@keunwoochoi

2 years

inference is straightforward. do the same with the training stage except - use MuLan text model, because we want *text*-to-music. - after SoundStream tokens are predicted, feed them to SS decoder to generated audio.

1

0

43

Keunwoo Choi

@keunwoochoi

2 years

it's official. #ismir2023 in Milano, Italy! 🎉

1

45

Keunwoo Choi

@keunwoochoi

5 years

“Academic conference in conputer science” or rather just a cult #ismir2019

2

3

44

Keunwoo Choi

@keunwoochoi

3 years

amazing, amazing. done by @ethanmanilow @pseetharaman et al.

AK

@_akhaliq

3 years

Unsupervised Source Separation By Steering Pretrained Music Models abs:

3

29

158

2

5

43

Keunwoo Choi

@keunwoochoi

3 years

Sheet Sage: Lead sheets from music audio Leverage Jukebox for melody extraction. Who'd submit this level of amazing work simply to late-breaking/demo session? This guy → @chrisdonahuey

1

6

43

Keunwoo Choi

@keunwoochoi

3 years

DrummerNet Gangs (K Choi and K Cho)

1

0

43

Keunwoo Choi

@keunwoochoi

2 years

#ISMIR 2022 Tutorials are all online! Help yourself to a cup of music AI :)

0

6

42

Keunwoo Choi

@keunwoochoi

1 year

🚨 We have a MLE position open at @PrescientDesign to find a strong engineer to make our language models stronger.

2

12

39

Keunwoo Choi

@keunwoochoi

1 year

Room Impulse Response Estimation in a Multiple Source Environment. By @kyungyunleee at @gaudiolab et al.

Room Impulse Response Estimation in a Multiple Source Environment

In real-world acoustic scenarios, there often are multiple sound sources present in a room. These sources are situated in various locations and produce sounds that reach the listener from multiple...

arxiv.org

3

2

40

Keunwoo Choi

@keunwoochoi

2 years

c4dm folks won the #ismir2022 best paper award!! 🎉🥳🎊 amazing! congrats, @liulelecherie @QiuqiangK @veromorfi @emmanouilb !

0

2

40

Keunwoo Choi

@keunwoochoi

4 years

Long time no first-authoring! Listen, Read, and Identify network (LRID-Net) identifies singing language by reading the metadata (title, album, artist) and listening to the audio.

Listen, Read, and Identify: Multimodal Singing Language...

We propose a multimodal singing language classification model that uses both audio content and textual metadata. LRID-Net, the proposed model, takes an audio signal and a language probability...

arxiv.org

1

4

38

Keunwoo Choi

@keunwoochoi

1 year

Our paper about DCASE Challenge T7 - Foley Sound Synthesis was accepted to the DCASE Workshop 🥳 I can't make it to Finland🇫🇮, but some of the authors will be there to tell you what we went through while organizing the first generative challenge at DCASE.

Foley Sound Synthesis at the DCASE 2023 Challenge

The addition of Foley sound effects during post-production is a common technique used to enhance the perceived acoustic properties of multimedia content. Traditionally, Foley sound has been...

arxiv.org

0

6

39

Keunwoo Choi

@keunwoochoi

3 years

GPT-3 is so 2020. saw it on the way @kchonyc ’s place. this must be a sign..

2

38

Keunwoo Choi

@keunwoochoi

4 years

ByteDance 🚀 US Speech / Audio / Music research team is extensively hiring research scientists. If you’re a graduating PhD this year, don’t wait and just DM me! 🔥🔥

3

12

37

Keunwoo Choi

@keunwoochoi

1 year

new music AI model alert 🚨 get your music tracks segmented by @taejun_kim_

Taejun Kim

@taejun_kim_

1 year

Music Structure Analyzer Released ✨ [Python Package] [Paper] [Interactive Demo] [Hugging Face Space]

11

55

260

2

3

38

Keunwoo Choi

@keunwoochoi

2 years

another day, another music generation paper! a diffusion one this time. i’m very curious where they got the training data 🤔

AK

@_akhaliq

2 years

Noise2Music, where a series of diffusion models is trained to generate high-quality 30-second music clips from text prompts project page:

11

83

493

2

0

38

Keunwoo Choi

@keunwoochoi

6 months

i'm giving an introductory talk about LLMs for drug discovery at #ASCPT2024 pre-conference soon.

2

8

37

Keunwoo Choi

@keunwoochoi

3 years

DawDreamer: A Python-interfaced DAW. Yeah we can do lot of things with this.

David Braun

@DoItRealTime

3 years

DawDreamer has gained many features recently including pip install. A new notebook shows how to load Ableton warp marker files like this video. Faust integration enables custom polyphonic instruments. Hopefully very useful for ML researchers and artists.

0

14

1

4

37

Keunwoo Choi

@keunwoochoi

2 months

✨ hiring an outstanding pm: - Prescient Design, Genentech - Challenging and novel LLM products in this domain - NYC first.last at gene dot com.

1

9

37

Keunwoo Choi

@keunwoochoi

5 months

teaching "deep learning for media" at NYU was super fun! now, let me disseminate my students' final projects. these are really cool stuff. they somehow made it in the vary last minute. i swear none of these was at this level just one week before 😂 anyways, 🧵 starts -

2

1

37

Keunwoo Choi

@keunwoochoi

3 months

looking for an enthusiastic MLE/SWE who *knows* LLMs and their deployment for our internal LLM serving through APIs and web-interfaces. ideally, 1-2yr industry experience + master grad level understanding in LLMs/ML. NYC. amazing team & great use-cases.

0

7

36

Keunwoo Choi

@keunwoochoi

3 years

Try it yourself our music source separation! 🚨 ALERT: The performance might be way too good.

Keunwoo Choi

@keunwoochoi

3 years

BTS - Dynamite, source separated by Gaudio™️

4

2

16

4

7

35

Keunwoo Choi

@keunwoochoi

5 years

New blog post - #ismir2019 poster bootlegs.

A bunch of ISMIR posters (or their bootlegs)

ISMIR 2019 was super great as usual – or even better. I, along with many other people, deeply appreciate Cynthia’s effort to make it more accessible and inclusive. The academic quality …

keunwoochoi.wordpress.com

1

8

36

Keunwoo Choi

@keunwoochoi

1 year

DCASE Task 7 - Foley Sound Synthesis has finished. It was the very first generative audio AI challenge. I'm very happy to have organized such a successful event! 🎉

1

2

35

Keunwoo Choi

@keunwoochoi

5 years

The longest ever video of me talking public has become public. "Deep Learning with Audio Signals: Prepare, Process, Design, Expect" in @QConAI . In case me tweeting around you isn't enough.

Deep Learning with Audio Signals: Prepare, Process, Design, Expect

Keunwoo Choi introduces what the audio/music research societies have discovered while playing with deep learning when it comes to audio classification and regression.

www.infoq.com

1

5

35

Keunwoo Choi

@keunwoochoi

3 years

import tensorflow as plt

2

1

35

Keunwoo Choi

@keunwoochoi

2 years

@rrherr And MuLan was already Google-verse only 😢

2

0

33

Keunwoo Choi

@keunwoochoi

11 months

look how shamelessly i'm included here! as always, it was great to connect to all the great researchers in MACLab supervised by @juhan_nam at @ISMIRConf .

Haven Kim

@havenpersona

11 months

This year, people from the Music and Audio Computing Lab at KAIST, led by @juhan_nam , participated in the @ISMIRConf , and presented our work through scientific programs, late-breaking demos and music sessions!

1

3

35

1

0

34

Keunwoo Choi

@keunwoochoi

10 months

Big news in AI this week - Mistral 7B on torrent - Google Gemini - and.. - my first single album <unspoken serenity> released;

keunwoo.OOO

keunwoo.ooo

2

0

34

Keunwoo Choi

@keunwoochoi

8 months

are you an LLM nerd who can understand ML/language model papers and write good code? 👀

Keunwoo Choi

@keunwoochoi

8 months

we're hiring AI/LLM engineers! - covering both pre-training and post-training tasks - purely for product development, based on *extensive understanding in LLMs* - with real-world impacts on drug discovery in Genentech - no publication within sight

1

12

54

2

4

32

Keunwoo Choi

@keunwoochoi

5 years

The #ismir2019 poster repo is hosting 25 posters and 38-starred now. Would you please 'Like' this tweet if you've ever been the repo and seen any posters there? I wanna know its impact. Thanks!

GitHub - keunwoochoi/ismir-2019-posters

Contribute to keunwoochoi/ismir-2019-posters development by creating an account on GitHub.

github.com

0

6

33

Keunwoo Choi

@keunwoochoi

7 months

generative AI audio is here to stay.. and prosper! check out this year's challenge. T7. Sound Scene Synthesis #DCASE2024

0

2

33

Keunwoo Choi

@keunwoochoi

6 years

최근우, 김사무엘(Samuel Kim) 및 174명의 음향, 음성 과학자는 PD수첩에서 방영한 숭실대학교 배명진 교수 사태와 관련 다음의 성명서를 발표합니다. 이 성명서는 한국음향학회 및 숭실대학교 전자정보공학부 임원진과 교수님들께도 발송하였습니다. 감사합니다. 성명서:

배명진 교수 사태에 대한 음향, 음성 과학자의 입장 [성명서].pdf

이 성명서는 음향/음성 학계의 신뢰 회복을 위한 공익적 목적을 위해 작성되었습니다.). (. 배명진 교수 사태에 대한 음향/음성 과학자들의 입장 년 5월 22일 방송된 ‘MBC PD수첩’(이하 PD수첩)은 소위 소리공학 전문가로 불리는 배명진 교수(이하 배 교수)의 활동과 그에 대한 의혹을 다루었다. 배 교수는 숭실대학교 교수와 ‘소리공학연구소’의...

www.docdroid.net

1

64

29

Keunwoo Choi

@keunwoochoi

6 years

After like 3 months of experiments (with some progress) I just realised out of N layers, good three of them didn't have an activation function at all.

5

1

32

Keunwoo Choi

@keunwoochoi

3 years

"Building the MetaMIDI Dataset: Linking Symbolic and Audio Musical Data" Hell a lot of midi files and matched audio clips. #ismir2021

1

4

31

Keunwoo Choi

@keunwoochoi

4 years

I've been an audio person for 10+ years. Let me tell you - you don't need 192/24 or anything. If you don't like the audio quality from any legit music streaming service, it's NOT about the codec. get a better connection, quieter place, better earbuds.

3

6

31

Keunwoo Choi

@keunwoochoi

2 years

oo more text-to-music to come. this time, from academia!

Haohe Liu

@LiuHaohe

2 years

Can't wait to share our new Text-to-Audio model, AudioLDM. 😆 This video shows the generation result with a simple text prompt: "A music made by xxx". More demos coming soon!😉 The paper will be available next Monday on arXiv! 😊 Our model will be open-sourced soon!😎

27

98

601

2

6

31

Keunwoo Choi

@keunwoochoi

6 years

Um, Spotify will definitely hire 2019 summer research interns for some fun MIR works, so please stay tuned! (i.e. don't say yes to others too soon 😎)

3

2

30