LALAL.AI’s innovative approach to stem separation is still the same: the user simply drags and drops the audio or video file on the page and the software outputs high-quality audio, separating the instrumental and acapellas from songs seamlessly.  Vocals and full instrumentals can still be isolated using LALAL.AI’s Casseiopia neural network that continuously improves its technology with every file upload.

Cassiopeia “is trained on 20TB of data to extract instrumentals and voice tracks from songs.” Three settings for processing files are now available – mild, normal and aggressive. Each setting determines how LALAL.AI processes the files, using its prior knowledge to scrub unwanted artifacts and digital remnants from each stem.

Three paid tiers are also available in LALAL.AI with more features, longer time and larger upload limits. The free tier allows 10 minutes of upload time; the entry-level Lite tier starts at $10 and provides 10 uploaded tracks or 90 minutes of audio. The $20 Professional plan (normally $30) provides 30 input tracks and 500 total minutes.  Business plan pricing is also available for larger data processing, starting at $1000/month for 10,000 minutes. 

The new instrument isolation update, like all of LALAL.AI’s isolation services, is available at no charge to any tier of user. Find more information and try out the AI-powered technology on the LALAL.AI website. 

In modern music production, mashups and remixes are the norm. Producers, DJs and even social media users are constantly creating unique edits of popular songs to showcase online. Several sites are available that use some form of AI technology to separate audio tracks, but the results return with unwanted audio artifacts. LALAL.AI’s system outputs high-quality audio every time, separating the instrumental and acapellas from songs almost seamlessly.

Separating tracks on LALAL.AI is simple – first drag or upload an audio file onto the page, then choose the processing level. After analyzing the track, the software offers previews of the separated vocal and instrumental. Users can then choose to download both as full tracks or try another process.

LALAL.AI operates using the Cassiopeia neural network, recognizing complex data patterns and mimicking how the human brain operates. Cassiopeia is an upgrade from LALAL.AI’s previous Rocknet environment, providing much deeper analysis in separating tracks. The LALA.AI website states Cassiopeia “is trained on 20TB of data to extract instrumentals and voice tracks from songs.” Previous Rocknet algorithm settings are also included as a rendering option, and provide more analysis – in some cases better than Cassiopeia.

LALAL.AI offers a free tier that limits the user to three tracks and a total of ten minutes. Three paid tiers are also available with more features and longer time limits. Lite, the entry-level tier, starts at $10 and provides 10 tracks and 90 minutes of audio. The $30 Professional plan provides 30 input tracks and 500 total minutes. For users looking to separate more files, a negotiable “High Volume” plan is available by contacting LALAL.AI directly.

Find more information and try out the AI-powered technology on LALAL.AI.