AIVoiceSeparator

AIVoiceSeparator is a web service and REST API that splits songs into vocal and instrumental stems. It is best for creators and developers making occasional separations who want flexible inputs without a subscription. Full Studio quality on the free plan is a strong starting point, but its monthly quota and single-GPU queue make frequent batch work a less natural fit.

Overview

The service produces backing tracks or clean vocal parts from a song. Its Studio pipeline combines BS-Roformer, Mel-Roformer and MDX23C models and applies EBU-R128 loudness normalization. The maker reports an SDR of 12.97 dB, about 3 dB above the open Demucs baseline. Treat that as a useful quality claim, not a guarantee across different recordings.

Privacy terms are specific: audio and results are automatically deleted within 24 hours, and uploaded audio is not used to train AI models. The maker says processing runs on its own GPU in Thailand, not a third-party cloud. Traffic is encrypted through Cloudflare HTTPS, and temporary files are stored on a private machine accessible only through Cloudflare Tunnel. For signed-in accounts, the service collects Google email and uid; it also uses IP addresses for rate limits and quota counting and collects request metadata for debugging, but does not log audio content. Users are responsible for having rights to the material they process.

Key features

The web app handles MP3, WAV, M4A, FLAC, OGG, WebM, Opus and MP4 or MOV video, with a 100 MB and 20-minute ceiling. YouTube and SoundCloud links can be submitted directly; TikTok clips must first be saved and uploaded. That range is convenient for varied sources, though the duration cap rules out longer recordings.

Optional Whisper transcription exports SRT, LRC and TXT files and adds about 30 seconds to processing. Free and Pro plans also include BPM and key detection. The REST API accepts either a URL or an uploaded file, authenticates with an X-API-Key and returns vocal and instrumental stems. Its RapidAPI listing and support for YouTube, SoundCloud, TikTok, Bandcamp and Vimeo links make it useful for developers, although batch throughput beyond the single-GPU queue requires contacting the maker about dedicated capacity.

Pricing

The freemium structure gives occasional users a way to avoid a recurring bill, while offering paid routes for additional web use or API workloads.

• Free: free, with three songs a month when signed in. It includes full Studio quality, MP3 320, WAV and FLAC exports, lyrics, BPM and key detection. The main trade-off is the small monthly allowance.
• Credits: $0.50, paid once, pay as you go with no subscription. Top-ups run from $1–$20 for 2–40 songs; credits never expire, there is no daily limit, and only successful separations are charged. This suits occasional users who exceed the free quota and do not want a monthly commitment.
• API credits: $0.50, paid once in prepaid credits, with no subscription or monthly minimum. Jobs up to seven minutes cost $0.50, jobs from 7–14 minutes cost $1.00, and jobs from 14–20 minutes cost $1.50; failed jobs are refunded. This usage-based schedule suits developers with variable demand, but costs rise with file duration.
• Pro: $30 a month, billed via Patreon and cancellable anytime. It allows six songs a day at the same Studio quality and with the same listed exports, lyrics, BPM and key detection as Free. The higher allowance is aimed at regular users, but the daily cap still makes it a poor fit for large batches.

The service also offers a free trial, though no trial length is given. Credits are the more flexible paid choice for sporadic work; Pro makes more sense for steady daily use.

Platforms

AIVoiceSeparator runs on the web and offers API access. Its web workflow supports audio and video uploads, while the API accepts URLs or files. No desktop or mobile app is part of the platform offering.

Who it's for

Choose it for occasional vocal removal, backing-track creation or stem delivery through an API, especially if a 20-minute limit and three free monthly songs are enough to start. Users who need recurring volume can pay per separation or move to Pro. Teams that depend on high batch throughput should ask about dedicated capacity before building a workflow around the standard queue.

Pros and cons

• Pro: The free plan retains full Studio quality and includes several export formats, lyrics, BPM and key detection.
• Pro: Non-expiring credits avoid a subscription and charge only for successful separations.
• Pro: Broad upload formats, video support and direct links cover more input routes than a file-only workflow.
• Con: The free allowance is limited to three songs per month when signed in.
• Con: Uploads stop at 100 MB or 20 minutes, and TikTok clips require a save-and-upload step.
• Con: Batch users may need dedicated capacity because the standard queue runs on a single GPU.

Alternatives

For a broader comparison, browse AI Voice Isolators.

• AudioShake is worth comparing if you need a wider range of platforms, including mobile, desktop and self-hosted options; its free plan starts with 200 credits and supports files up to 10 minutes.
• LALAL.AI may suit users who prefer an always-free relaxed queue with a 10-minute allowance and 200 MB upload limit, though its free plan offers previews rather than full downloads.
• Altered Real-Time Pro is the more relevant alternative for desktop speech enhancement and voice skins, with a free daily processing allowance on macOS and Windows.
• VocalDrop is an option for desktop users seeking a free, no-account workflow with no track limits or watermark.
• Demixr may fit users who want a web-based alternative with a $4 a month Go plan, but its free tier limits monthly audio and upload size and does not provide WAV downloads.
• Denoise is a mobile option for denoising audio on Android or iOS; saving and exporting requires Pro.
• Vocly is an Android alternative with a free tier that includes ads and karaoke mode.
• clir Focus is an iOS option that starts with three free sessions.

Verdict

AIVoiceSeparator is a sensible choice for creators and developers who want full-quality occasional separation, multiple input routes and credits that do not expire. Its clearest advantage is avoiding a subscription for intermittent work; its clearest drawback is limited capacity for users who need frequent or high-volume batches. Start with Free, use credits for uneven demand, and look elsewhere if throughput is the priority.