A new kind of sampler: sonic experimentation with AudioStellar's AI generated sound map
Workshop Title
A new kind of sampler: sonic experimentation with AudioStellar’s AI generated sound map
Description
Along with our paper “AudioStellar, an open source corpus-based musical instrument for latent sound structure discovery and sonic experimentation” submitted for ICMC2020 we would like to offer a hands-on workshop that will demonstrate the main concepts and features of the AudioStellar software.
AudioStellar is a sampler-like musical instrument that provides a user interface for visualizing a collection of short audio clips in an interactive 2D point map generated by an AI pipeline. This sound map can be played using creative techniques that are difficult to achieve using traditional DAWs or even custom code.
Audiostellar was developed by a team of Argentinian programmers and debuted at NIME 2019. Since its first demonstration, Audiostellar has received a lot of attention from users across the globe--in January 2021, Audiostellar’s website had 200 downloads and 1000 visits from more than 20 countries in 4 continents, including the US, Germany, Japan, India, and, of course, Argentina.
The workshop will cover all the features the software offers for playing, composing and sound designing, including: the different types of units and effects, drawing trajectories on latent space, creating new sound maps, using MIDI, OSC and external hardware for controlling the software and creating custom behaviors using examples in Max and TouchOSC. Although this workshop will have an artistic and musical approach, basic notions of unsupervised machine learning, focusing on dimensionality reduction, clustering and intuitions on how Audiostellar’s AI pipeline works, will be introduced to the attendees.
After this introduction, an exhaustive explanation will cover the different tools that AudioStellar can offer to the diverse users. The 4 different types of Units inside Audiostellar will be explained in detail, as long as the parameters included in them. Moreover, the particularities in clustering audio samples as well as the importance of visualization of data in music will be discussed. The workshop has also the intention to generate an exchange of opinions related to the musical and artistic approaches that can arise from the use of AudioStellar’s tools when creating or performing a music composition. This will be considered as an important instance of the workshop, as the feedback between users may extend the uses of AudioStellar. Additionally, examples of connectivity between AudioStellar and external software and hardware such as Max or devices using TouchOSC will be demonstrated.
The activity will finish with hands-on experience where participants will use the software in their own notebooks and we will answer the specific questions that might arise. We encourage participants to bring their own audio samples for experimentation, as diversity in audio samples will give more perspectives to the whole atten. If it is not possible, folders of audio samples will be sheared.
This tutorial is addressed to sound artists, sound designers, musicians, instrument makers, audio enthusiasts and creative practitioners in general. AudioStellar is free, open source and available for Linux, Mac and Windows. More info at: http://www.audiostellar.xyz
Short Description
We will offer a hands-on workshop to share the main concepts and features of AudioStellar (open source) for musical and artistic purposes. AudioStellar provides a user interface for visualizing a collection of audio clips in an interactive 2D point map generated by an AI pipeline as well as a group of tools that make it possible to play them using creative techniques. Our aim is to introduce the software to the participants, but beyond the introduction, we expect the attendees to experience the software, discover the possibilities that the software may offer as well as answer the specific questions that might arise.
Organizers
List the Workshop organizer(s). For each organizer, include:
Agustín Spinetto MA
Tokyo University Of the Arts
agustinspinetto@gmail.comAgustín Spinetto is a musician from Argentina. He has a bachelor’s degree in Electronic Arts from the UNTREF University, Buenos Aires, where he has also taught courses in sound recording. In 2018, Agustín was awarded the MEXT Scholarship by the Government of Japan and moved to Tokyo for graduate work. He received a master's degree in Music and Sound Creation from the Tokyo University of the Arts in March 2021. During his time in Tokyo, Agustin has researched the use of visual and digital interfaces for music performance and his master’s thesis focused on controlling analog synthesizers using Audiostellar.
Since his time as undergraduate he has been working with electronic and acoustic music instruments and using new technologies for music experimentation purposes. His studies led him to work with music improvisation using synthesizers and music programming languages, while also engaging with more popular music genres such as rap, hip hop, jazz and noise rock. Throughout his career he has performed in several countries around the globe, such as the USA, Japan, South Korea and Argentina where he performed using AudioStellar.Lic. Leandro Garber
MUNTREF Art and Science Center
leandrogarber@gmail.comLeandro Garber is an artist, teacher, programmer and data scientist from Buenos Aires. He studied Computer Science at the “Universidad de Buenos Aires (UBA)”, has a degree in Electronic Arts from the “Universidad de Tres de Febrero (UNTREF)”, and is writing his thesis on speech language identification as part of a master degree in Data Science at UBA. He teaches creative coding at UNTREF where he also coordinates a research group focused on generative and explorative machine learning algorithms. He is a data scientist at CIIPME-CONICET where he is part of a research group studying child language development.
Preferred Length of Workshop
Half day (4 hours)
Technical and Space Requirements
This workshop could be done in-person, virtual or blended formats.
The workshop will required:
A projector with a HDMI cable
A pair of speakers (with ⅛ stereo jack cable or 2 ¼ mono jack cables)
Internet connection
Each participant of the workshop will need:
A notebook with a working internet connection
A copy of Audiostellar already installed in their host operating system (Participants with installation issues will be helped if needed).
A folder of sound files to build their own sound maps (ideally, more than 100 files) (A default folder of audio samples will be provided if needed)
A pair of headphones
Optional links to supporting audio, visual, and/or online documentation, e.g. a workshop web page. Any media materials should use an unrestricted streaming service such as Vimeo or Bilibili.
Presentation tools (if online due COVID)
Expository section (2 hours): YouTube live stream
Practical section (2 hours): Zoom
Links to Supporting Media (optional)
www.audiostellar.xyz <— Audiostellar website
http://icmc2021.org/selected-papers/ <— Link to check AudioStellar paper on ICMC2020
https://www.youtube.com/channel/UC_p3ffu7ceFNlapBGWx7GhA <— AudioStellar channel on youtube
https://youtu.be/2YxONrfA6po <— Performance done with AudioStellar presented in NYCEMF 2020.