Stability AI is taking its generative AI tech into the world of music because the developer has launched a brand new text-to-audio engine known as Secure Audio.
Much like the Secure Diffusion mannequin, Secure Audio can create brief sound bites based mostly on a easy textual content immediate. The corporate explains in its announcement publish that the AI was educated on content material from the net music library AudioSparx. It even claims the mannequin is able to creating “high-quality, 44.1 kHz music for industrial use”. To place that quantity into perspective, 44.1 kHz is taken into account to be CD high quality audio. So it’s fairly good however not the best.
A free model of Secure Audio is at the moment out there to the general public the place you’re allowed to generate and obtain 20 particular person tracks a month. Every sound chunk has a forty five second runtime so that they received’t be very lengthy.
Prompting music
The textual content prompts you enter could be easy inputs. Listening to the samples supplied by Stability AI, “Automotive Passing By” sounds precisely because the title suggests – a automotive driving by within the distance though it’s a little muffled. Conversely, you can even stack on particulars. One specific pattern has a immediate involving Ambient Techno, an 808 drum machine, claps, a synthesizer, the phrase “ethereal”, 122 BPM, and a “Scandinavian Forest” (no matter which means). The results of this phrase mixture is an ambient lo-fi hip-hop beat.
We took Secure Audio out for a fast spin. We have been in a position to enter one immediate asking the AI to create a fast-paced storage rock tune from the early 2000s and it form of completed the objective. The generated monitor matched the model though it sounded actually messy.
Sadly, we couldn’t go any additional moreover the one enter. On the time of this writing, Secure Audio is seeing an enormous inflow of visitors from folks dashing in to check out the mannequin. The developer recommends attempting once more later or the subsequent day when you’re met with nothing however a clean display.
There’s a catch with the free model – it’s for non-commercial use solely. If you wish to use the content material commercially, you then’ll should buy the $12 Secure Audio Skilled month-to-month plan. It additionally gives 500 monitor generations a month, every with a length of as much as 90 seconds. There’s an Enterprise plan too for customized audio length and month-to-month generations. You’ll, nonetheless, should contact Stability AI first to arrange a plan.
Do bear in mind the expertise isn’t excellent. The content material sounds effective for essentially the most half, nonetheless sure points will appear off. The combo in that Ambient Techno tune talked about earlier isn’t excellent in our opinion. It was just like the bass and synthesizer are combating over what would be the dominant sound, leading to simply noise. Moreover, it doesn’t seem the AI can do vocals. It solely does instrumentals.
Secure Audio is fascinating for positive, however not one thing that needs to be completely relied on. We must always observe the corporate is asking for suggestions from customers on find out how to enhance the AI. A contact e mail can discovered on the official announcement web page.
In case you plan on using this tech on your personal function, we suggest checking TechRadar’s record of the greatest audio editors for 2023 to repair any flaw you would possibly come throughout.
