Recently, "Your Steel Gates Are More Rock" has swept the internet, and behind its success lies the driving force of Suno AI technology. Following Suno's immense popularity, Stability AI, the parent company of AI drawing application Stable Diffusion, has introduced an upgraded version of its music composition tool — Stable Audio 2.0 (hereafter referred to as SA2). The new version allows users to generate music up to 3 minutes in length, surpassing Suno by an extra minute!
Upon registration, SA2 users receive 20 points which can be used to generate music. The point consumption is not affected by the duration of the audio. Free users can generate up to 20 pieces of music per month, which is significantly less than before the restrictions were imposed on Suno.
On the SA2 generation page, users can generate personalized music in just 2-3 steps. First, in the prompt box at the top left corner, enter relevant requirements such as genre, instruments, emotional terms, and the BPM (beats per minute) tempo.
Next, users can customize their choice of model generation, with the default being the SA2.0 model. Each use of the SA2.0 model to generate audio consumes 2 points (up to 3 minutes), while the SA1.0 model consumes 1 point (up to 1 minute and 30 seconds). For free users who exclusively use the 2.0 model, they can actually generate a maximum of only 10 audio clips per month.
Finally, adjust the desired music duration, then click the "Generate" button to create music from scratch with no prior experience needed.
The major upgrade this time is the addition of audio-to-audio generation function in SA2, which allows users to regenerate audio samples using cue words. For example, users only need to record and upload a hummed vocal sample with cue words to get a melody played by an instrument.
Even though Stable Audio has iterated to version 2.0, when compared to Suno AI, everything becomes subtle. On one hand, SA2 leans more towards generating pure music, with the generated vocal tracks sounding incomplete and heavily skewed towards electronic sounds, almost like a tone-deaf recording played in reverse. If we were to describe the experience, it would be enough to make you laugh yourself to death.
On the other hand, most generated music lacks a strong sense of rhythm, with melodies that tend to be too straightforward and lacking in surprise. Additionally, the inability to generate Chinese tracks is a significant drawback for Chinese users.
Overall, the birth of SA2 is a cause for celebration. While there is still room for improvement in its ability to generate lifelike music, its prowess in audio-to-audio conversion has fulfilled the expectations of many music enthusiasts for exploration. Firstly, it has enhanced the efficiency of timbre conversion, and secondly, it has paved the way for many music creators, thus enriching another track in the AI audio domain.
The evolution of AIGC has been nothing short of rapid, from basic chatbots to the production of audiovisual works today. With AI tools at your disposal, you're essentially a one-person team! However, AI production always comes back to one key element: having a high-performance graphics card. Enter the Colorful RTX 4070 Ti SUPER Master OC, boasting a whopping 16GB of VRAM to effortlessly meet the demands of various AI applications. Its revolutionary TensorRT acceleration can also significantly enhance your productivity, transforming you into an efficiency powerhouse!
