Music Cognition and Informatics
411 views | +0 today
Follow
Music Cognition and Informatics
how does listening work? how can we model it?
Curated by Eric Nichols
Your new post is loading...
Scooped by Eric Nichols
Scoop.it!

Popular YouTube artist uses AI to record new album

Popular YouTube artist uses AI to record new album | Music Cognition and Informatics | Scoop.it
American Idol alum Taryn Southern is using AI as a part of her songwriting process.
Eric Nichols's insight:
"Southern has only basic piano skills, so she turned to the program to deliver the instrumental part of the song. The AI developed the harmonies, chords and sequences."

'In a funny way, I have a new song-writing partner who doesn't get tired and has this endless knowledge of music making,' Southern told CNN Tech. 'But I feel like I get to own my vision; I iterate and choose what I like and don't like. There's a lot more control.'"
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Machine learning lets computer create melodies to fit any lyrics

Machine learning lets computer create melodies to fit any lyrics | Music Cognition and Informatics | Scoop.it

"AI lends a hand to songwriters by setting their words to a pop song melody. And later, the aim is to get it to create whole compositions with lyrics by itself"

Eric Nichols's insight:
This actually cites a paper I wrote, and uses a formula and some ideas from my paper on rhythm generation.
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

WaveNet: A Generative Model for Raw Audio | DeepMind

This post presents WaveNet, a deep generative model of raw audio waveforms. We show that WaveNets are able to generate speech which mimics any human voice and which sounds more natural than the best existing Text-to-Speech systems, reducing the gap with human performance by over 50%.

We also demonstrate that the same network can be used to synthesize other audio signals such as music, and present some striking samples of automatically generated piano pieces.
Eric Nichols's insight:
WaveNet does a better job than I expected at synthesizing speech and even (random) music. Check out the audio examples.The impressive thing is that it works from raw 16kHz audio.
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Baidu AI Composer creates music inspired by art

Baidu AI Composer creates music inspired by art | Music Cognition and Informatics | Scoop.it

Not sure what to think of this -- flashy demo, but it's unclear how they generated the music. I suspect a lot of human selection in the mix to get to these examples. I did find it interesting that they apparently use song lyrics to help build their training set.

Eric Nichols's insight:
Share your insight
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

The Bitcoin Blockchain Just Might Save The Music Industry...If Only We Could ... - Forbes

The Bitcoin Blockchain Just Might Save The Music Industry...If Only We Could ... - Forbes | Music Cognition and Informatics | Scoop.it
The Bitcoin Blockchain could save the music industry and help artists monetize and track their creations. However, failure to understand this technology means it may go the way of Creative Commons.
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

The Year in Social and Streaming Charts: Katy Perry and Justin Bieber Rule

The Year in Social and Streaming Charts: Katy Perry and Justin Bieber Rule | Music Cognition and Informatics | Scoop.it
Streaming and on-demand services continued to play a significant role in determining a song’s popularity on the Billboard Hot 100 and other charts
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Large-Scale Pattern Discovery in Music

This is a quick overview of my Ph.D. thesis. It tries to answer: what is about? and, is it worth it for you to read it? You can get the PDF here. Quick background: I defended my thesis in January &...
Eric Nichols's insight:

From Thierry Bertin-Mahieux: an overview of his thesis work. Good stuff on analysis of analysis of large audio datasets.

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Melody Extraction

Melody Extraction | Music Cognition and Informatics | Scoop.it

Looks like a nice description of the authors' technique, and there's a plug-in to try it out.

 

"What is melody extraction? How is it performed? What applications is it good for? Demos, Vamp plug-in and further information related to melody extraction!"

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

The Pitch-Height Metaphor -- David Huron

Why do we call high pitches "high" and low pitches "low"?
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Analysis of chord transitions in 1300 popular songs

Analysis of chord transitions in 1300 popular songs | Music Cognition and Informatics | Scoop.it

Nothing too surprising here, but it's a nice simple intro to the topic of chord transitions in popular music. Earlier I mentioned David Temperley's analysis of the IV chord.  And I have my own paper based on my internship with Microsoft Research, where we looked at chord transitions in different genres: http://ericpnichols.com/papers/iui2009.pdf

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Daniel Levitin: Why Music Moves Us

You know the feeling. You hear "that song" and it evokes a certain emotion or memory. Cognitive psychologist Daniel Levitin sits down with Steve Paikin to ex...
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Music and the Ties that Bind | Psychology Today

An interesting article about a study on something I always thought was true: you're more likely to like people who share your taste in music, because, the article argues, music taste is linked to our core values.

 

"So whether your musical tastes run to the music of Nina Simone, Jean Sibelius, or Jay-Z, finding someone who shares those tastes might just be the beginning of a beautiful relationship."

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Get rhythm: Why the key to finding music you like is rhythm, not genre

Get rhythm: Why the key to finding music you like is rhythm, not genre | Music Cognition and Informatics | Scoop.it
So close and yet so wrong -- you might love heavy metal like Metallica but your music platform suggests you should also like the '60s sound of the Doors, simply because both bands are classified as rock.
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

AI and machine learning will make everyone a musician

AI and machine learning will make everyone a musician | Music Cognition and Informatics | Scoop.it
Sónar+D 2017: How AI and machine learning are shaping the future of music
Eric Nichols's insight:
"The move to AI-based music creation tools will be "as big a technological shift as the digitisation of music," he [Jon Eades] predicted, albeit cautiously.

"The Magenta project has been running for just over a year and aims discover whether machine learning can create "compelling" creative works. "Our research is focused on sequence generation," Eck says, “we’re always looking to build models that can listen to what musicians are doing. From that we can extend a piece of music that a musician’s created or maybe add a voice"."
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Universal's Audible Watermark

Eric Nichols's insight:
Music tracks from Universal apparently would sound bad even with lossless compression -- they introduce an audible artifact to "watermark" their music.
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Magenta

Magenta is a project devoted to music and art generation with machine intelligence. It is part of TensorFlow, an open source machine learning library.Magenta is a framework intended for

Eric Nichols's insight:
Magenta is a framework intended for generating music (and art) with TensorFlow -- focus seems to be on using RNNs/LSTMs for melody generation.
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Fundamentals of Music Processing: Audio, Analysis, Algorithms, Applications: Meinard Müller: 9783319219448: Amazon.com: Books

Fundamentals of Music Processing: Audio, Analysis, Algorithms, Applications [Meinard Müller] on Amazon.com. *FREE* shipping on qualifying offers. This textbook provides both profound technological knowledge and a comprehensive treatment of essential topics in music processing and music information retrieval. Including numerous examples
Eric Nichols's insight:

Looks like a great text.

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Crowdsourcing lets the masses compose song one note at a time

Crowdsourcing lets the masses compose song one note at a time | Music Cognition and Informatics | Scoop.it
A new site invites people to vote on which note should come next in a melody. Can a mob of strangers on the internet really write a good tune together?
Eric Nichols's insight:

I have some criticisms of how this was implemented (especially the repetitive chord progression and an apparent bias towards an isochronous rhythm), and I agree with Dave Cope's comments, but it's fun in any case. Vote here: https://crowdsound.net/

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

From James Taylor to Taylor Swift: Music evolves like biological organisms

From James Taylor to Taylor Swift: Music evolves like biological organisms | Music Cognition and Informatics | Scoop.it
Findings come thanks to largest data-driven study of pop music ever undertaken
more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Gangnam Style: We need more foreign language pop songs.

Gangnam Style: We need more foreign language pop songs. | Music Cognition and Informatics | Scoop.it

Thoughtful article about foreign-language pop. Some excerpts:

 

"How to account for the more than 650 million YouTube views of “Gangnam Style”? That jaunty dance surely deserves some credit, but might the faucalized voice and aspirated consonants of the Korean language play a part as well?"

 

"English-only listening habits deprive us of the natural rhythm and melody of other languages—the nasal vowels of French, the alveolar trills of Portuguese, the consonant clusters of Czech. That most of us don’t understand the words only allows us to better appreciate the phonology of a language and concentrate on the human voice as a musical instrument."

 

"A 2002 study in the cognitive science journal Cognition found that a culture’s dominant language influences the rhythmic structures of its instrumental music. In a 2006 study published in Psychology of Learning and Motivation, McGill University researchers write that Spanish features a “regular beat pattern in which each syllable coincides approximately with a beat, whereas stress-timed languages like English tend to have beats on stressed syllables.” In other words, languages each have their own rhythm, so certain languages tend to pair better with certain rhythmic patterns."

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

New book: Computers and Creativity

New book: Computers and Creativity | Music Cognition and Informatics | Scoop.it

I just ordered the last copy available new on Amazon... it's pricey... hope that it's worth it. It's an edited volume of papers on computing and creativity, and I recognize some names of good researchers in the table of contents. Most importantly, there is a large section on music, so I'm particulary interested in and have high hopes for those chapters.

more...
Christian Sarti's comment, May 30, 2013 6:12 PM
Hi, Eric Nichols.Do you know where can I find this book on-line? There's a chapter I've been trying to download: Chapter 6, "live algorithms: Towards autonomous computers improvisers". thank you very much.
Christian Sarti's comment, May 30, 2013 6:12 PM
Hi, Eric Nichols.Do you know where can I find this book on-line? There's a chapter I've been trying to download: Chapter 6, "live algorithms: Towards autonomous computers improvisers". thank you very much.
Scooped by Eric Nichols
Scoop.it!

Measuring the Evolution of Contemporary Western Popular Music : Scientific Reports : Nature Publishing Group

Measuring the Evolution of Contemporary Western Popular Music : Scientific Reports : Nature Publishing Group | Music Cognition and Informatics | Scoop.it

Congrats to this group for this feature article in Nature. A few things frustrate me though -- the really crucial stuff that's important for music is missing from the analysis: individual notes with their pitches and rhythms.  

 

Not to blame the researchers, or the people who put together the "Million Song Dataset". MSD, as great and useful as it is, doesn't have actual score-like data about the notes of a song -- individual pitches and rhythms are missing. After all, pitch transcription is an unsolved problem in the field. Instead, the MSD uses "chroma vectors" for each beat, which can be used (albeit imperfectly) for harmonic analysis. I'm worried that people will misinterpret the music notation at the top of Figure 1 of the article and think that individual note events were used in this research -- they weren't. Also, I think the term "pitch transition" in the article might be misleading; it's these crazy chroma vectors that are having transitions, not individual pitches.

 

One day we'll have databases that include symbolic score information, and it will be extremely useful. After all, Google indexes the web well because it has a collection of words in text to analyze... imagine if every web page were replaced by a recording of a person reading the contents and we deleted all the text; Google would be much less effective.

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Iamus, a computer composition program

Iamus, a computer composition program | Music Cognition and Informatics | Scoop.it

In this article in The Guardian, you can take a "Musical Turing Test" and try to guess which program was compsed by computer.  I've done that for Dave Cope's program EMI, but in this case it's the program Iamus, from the group in Malaga, Spain.

 

I listened to the 5 choices, picked the obvious computer-composed excerpt, and got it right. I'm a little surprised that there were so many incorrect votes, but part of it is the test format -- it presents a variety of styles, and it's a little confusing comparing apples and oranges. Also, contemporary classical music often sounds rather undecipherable and computer-like; I can see why it's hard to tell if random notes with no obvious structure to the blips and tone clusters was generated by a crazy composer or a random number generator. I hope that in future tests like this people will pick a more homogenous set of examples to chose from.

 

Take the test yourself! The link is in the article.

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

A Corpus Analysis of Rock Harmony

Interesting new work by Trevor de Clercq and David Temperley. The PDF is available on Temperley's website (http://theory.esm.rochester.edu/rock_corpus/declercq-temperley-pm11.pdf) I had suspected many of their findings, but it's great to see them spelled out in detail, backed up by statistics. For example, the IV chord is used heavily and in a different way than in classical music, and progressions don't seem to have much directionality in rock: IV can go to V and IV can go to V, and it doesn't really matter. I think a lot of what I don't like about some rock is due to this difference in harmony; I suspect that the rock/pop music I do like actually follows classical rules more than in the average Billboard 500 hits used in this study.

 

From the conclusions of the paper:

-----

The distribution of relative roots in the corpus shows a clear hierarchical pattern.
The much greater frequency of IV and V over other chords gives these chords a privileged position among non-tonic harmonies. IV is, however, noticeably more frequent than V, suggesting that it has a unique and primary status. After IV and V, VI and bVII form a clear secondary category; the next most common harmony (II) is only about half as common as these. With regard to patterns of harmonic progression, the strong asymmetries of root motion found in common-practice music are notably absent in rock. The frequencies of chords in post-tonic and pre-tonic position are, for the most part, similar to their overall frequencies. Some positional preferences do, however, emerge. Most significantly, the frequency of IV as a pre-tonic harmony is somewhat higher than its overall frequency, suggesting that it often functions to prepare the arrival of the tonic in some way...

more...
No comment yet.
Scooped by Eric Nichols
Scoop.it!

Of earthquakes, hurricanes, the cloud, music, hacking and other forms of destruction

Of earthquakes, hurricanes, the cloud, music, hacking and other forms of destruction | Music Cognition and Informatics | Scoop.it
Of earthquakes, hurricanes, the cloud, music, hacking and other forms of destruction http://t.co/TNwm5kb...
more...
No comment yet.