gkucsko

7 ms·

1.
Suno AI and Timbaland collaborate on new music(vibe.com)
1 points by gkucsko 2y ago|0 comments
2.
Grammy winner Timbaland drops new single exclusively on Suno(timbaland.suno.com)
2 points by gkucsko 2y ago|0 comments
3.
Moshi: A speech-text foundation model for real time dialogue(github.com)
365 points by gkucsko 2y ago|64 comments
4.
by gkucsko 3y ago
Our models support full 2+ mins of coherent generation but generating a couple of verses at a time through continue gives good results you can keep picking the continuations that sound best!
5.
by gkucsko 3y ago
We’re US based but come from pretty much across the globe and my wife is Punjabi, so yeah that’s the origin of the name :)
6.
Show HN: Lyrics-Prompted AI Music Generation(twitter.com)
1 points by gkucsko 3y ago|0 comments
7.
Suno VOL-1: Generative Music with Vocals(suno-ai.notion.site)
28 points by gkucsko 3y ago|3 comments
8.
by gkucsko 3y ago
yeah sometimes there are definitely artifacts. technically they can be removed pretty easily with another model (like denoiser from FB) but for now we wanted to keep it simple to learn to control these things better through prompt engineeri
9.
by gkucsko 3y ago
history is semantic, coarse and fine. so essentially the same thing thats getting generated just using it as an input before the generation
10.
by gkucsko 3y ago
thanks, the model itself is a pretty vanilla gpt model based heavily on karpathy's nanogpt, so should not need too many bells and whistles to get it running on specific architectures. that said i have very little experience with platfo
11.
by gkucsko 3y ago
Hey, one of the Suno founders/creators of Bark here. Thanks for all the comments, we love seeing how we can improve things in the future. At Suno we work on audio foundation models, creating speech, music, sounds effects etc…. Text to
12.
by gkucsko 3y ago
haha https://demo.suno.ai
13.
by gkucsko 3y ago
history prompts are just unconditionally generated TTS from the same model. any of those can be used as history, but for convenience 10 are provided for each language (to generate things with consistent voices)
14.
by gkucsko 3y ago
it's more meant to show code switching. more examples here: https://suno-ai.notion.site/Bark-Examples-5edae8b02a604b54a4...
15.
by gkucsko 3y ago
Thanks :) Let us know if you come across interesting findings!
16.
by gkucsko 3y ago
Thanks! The model learns a lot from unsupervised (as well as supervised) audio, so technically low-quality and high-quality audio are both just as likely to the model as music, background sounds or really anything else including echos or ba
17.
GPT-4-Audio: generative Text-to-Audio using Bark(github.com)
16 points by gkucsko 3y ago|4 comments
18.
SPGISpeech: 5k hours of transcribed audio for ASR(datasets.kensho.com)
17 points by gkucsko 5y ago|0 comments
19.
by gkucsko 12y ago
There is no sign-up (microsoft account or similar) necessary, the broadcast link is much shorter and easier to share, and it's open source :)