Abstract We introduce MusicLM, a model generating high-fidelity music from text descriptions such as "a calming violin melody backed by a distorted guitar riff". MusicLM casts the process of conditional music generation as a hierarchical sequence-to-sequence modeling task, and it generates music at 24 kHz that remains consistent over several minutes. Our experiments show that MusicLM outperforms previous systems both in audio quality and adherence to the text description. Moreover, we demonstrate that MusicLM can be conditioned on both text and a melody in that it can transform whistled and hummed melodies according to the style described in a text caption. To support future research, we publicly release MusicCaps, a dataset composed of 5.5k music-text pairs, with rich text descriptions provided by human experts.
So you're all done recording your next song. You've laid down final takes for all the tracks, mixed everything and decided on the final master. Congrats!
But before you call it a day and prepare to distribute to streaming platforms, there are a few things many musicians forget to do that can take their song to the next level.
Audio on Unix is a little zoo, there are so many acronyms for projects and APIs that it's easy to get lost. Let's tackle that issue! Most articles are confusing because they either use audio technical jargon, or because they barely scratch the surface and leave people clueless. A little knowledge can be dangerous.
In this article I'll try to bridge the gap by not requiring any prerequisite knowledge while also giving a good overview of the whole Unix audio landscape. There's going to be enough details to remove mysticism (Oh so pernicious in web bubbles) and see how the pieces fit.
The SOUL project is creating a new language and infrastructure for writing and deploying audio code. It aims to unlock improvements in latency, performance, portability and ease-of-development that aren't possible with the current mainstream techniques that are being used.
Midi Quest supports over 850 of the most popular MIDI hardware and instruments from over 50 different manufacturers including Korg, Roland, Yamaha, Dave Smith, Kurzweil, Alesis, Waldorf, Kawai, Akai, and E-mu.
You can store, organize, and edit banks and the individual patches, combinations, multis, performances, drums settings, and other SysEx data loaded from your MIDI hardware. Midi Quest is a true multi-instrument editor/librarian designed from the ground up to effectively support multiple MIDI ports, multiple manufacturers, and multiple MIDI devices - including multiples of the same hardware.
With a rich network of sound-obsessed cafés, bars and small clubs, Aaron Coultate explains why Tokyo might be the best place in the world to listen to music.
I've heard somewhere that you're not an author until you've had two books published. If this rule also applies to music then I guess I am now officially a musician, or something. Wretched Saints, the music project I am involved with, now has its second single out. It's called "I Am Become Death" and you should be able to find it here:
Meanwhile, we're working on our next song, arguing over silly design details on our website and trying to figure out what all those audio mixing knobs do and how they can make music sound better.
I have just learned that a song from my music side project Wretched Saints, which doesn't even have a website yet, is now out on Spotify:
If you're not a Spotify user, here's an older version of this song on Soundcloud:
This is the jazz piano site of Doug McKenzie. It contains many downloadable video files in WMV format and midi files of live played songs.
Control any MIDI enabled hardware: syntesizers, drum machines, samplers, effects.
Create custom interfaces.
Host them as VST or AU plugins in your favorite DAWs.