AIVoiceSeparator

AIVoiceSeparator is a web app and REST API that splits songs into vocal and instrumental stems, with optional lyric files. It is best for creators who need occasional, high-quality separation and developers who want URL or upload-based automation. The main trade-off is a small free allowance and a single-GPU queue that may not suit sustained batch work.

Overview

The service produces separate vocals and instrumentals for backing tracks, clean vocal parts and similar editing tasks. Its Studio pipeline combines BS-Roformer, Mel-Roformer and MDX23C, then applies EBU-R128 loudness normalization. The maker reports an SDR of 12.97 dB, approximately 3 dB above the open Demucs baseline. That is a useful quality indication, though source material will still affect results.

Privacy handling is unusually concrete. Processing runs on the maker's own GPU in Thailand rather than a third-party cloud; uploaded audio and results are deleted within 24 hours, and audio is not used to train AI models. Traffic uses Cloudflare HTTPS, with temporary files on a private machine reachable only through Cloudflare Tunnel. Signed-in accounts provide Google email and uid; IP addresses support rate limiting and quota counting, and request metadata is retained for debugging without logging audio content. Users are responsible for rights to material they upload.

Key features

The web uploader handles MP3, WAV, M4A, FLAC, OGG, WebM, Opus and MP4/MOV video, subject to a 100 MB and 20-minute ceiling. YouTube and SoundCloud URLs can be submitted directly. TikTok clips must first be saved and then uploaded, adding a manual step. These inputs cover common creator workflows, but a recording beyond either limit needs another tool or preprocessing.

Whisper transcription is optional and returns SRT, LRC and TXT files, adding about 30 seconds to processing. BPM and key detection are included on the Free and Pro plans, making those tiers more useful for arrangement and practice workflows than a stem-only service.

The REST API accepts either a URL or an uploaded file, authenticates with an X-API-Key and returns vocal and instrumental stems. It is listed on RapidAPI and accepts YouTube, SoundCloud, TikTok, Bandcamp and Vimeo links. Developers can therefore integrate multiple media sources without building separate download handling. For workloads beyond the single-GPU queue, the API page directs customers to the maker for dedicated capacity, so larger operations should confirm throughput before committing.

Pricing

Free costs nothing and allows three songs per month when signed in. It retains full Studio quality and includes MP3 320, WAV and FLAC exports, lyrics, BPM and key detection. This is an excellent evaluation tier for occasional users, but three songs will not cover regular production or a sizable back catalog.

Credits cost $0.50 as a one-time purchase, billed pay as you go with no subscription. Top-ups run from $1 to $20 for two to 40 songs; credits never expire, there is no daily limit, and payment applies only to successful separations. This is the sensible choice for irregular demand because unused capacity does not vanish, although frequent users may find repeated top-ups less predictable than a monthly plan.

API credits also cost $0.50 as a one-time purchase as prepaid credits, with no subscription or monthly minimum. Jobs up to seven minutes cost $0.50, seven to 14 minutes cost $1.00, and 14 to 20 minutes cost $1.50; failed jobs are refunded. The duration bands make costs easy to estimate, while longer tracks consume credits quickly.

Pro costs $30 a month, billed via Patreon, and can be cancelled anytime. It permits six songs per day with the same Studio quality, MP3 320/WAV/FLAC output, lyrics, BPM and key detection as Free. Pro suits creators who regularly approach the free cap; its daily ceiling still limits heavy batch production.

Platforms

AIVoiceSeparator runs on the web and through an API. Input supports both audio and video, with upload processing, a 20-minute maximum duration and MP3, WAV and FLAC exports. There is no stated native desktop or mobile application, so browser and integration users are the intended audience.

Who it's for

Choose it for a singer, producer or DJ needing a few full-quality separations each month, especially when lyric files and BPM or key data matter. The prepaid credit model also fits occasional catalog work, while API credits suit developers automating media ingestion. Teams with continuous batch requirements should discuss dedicated capacity first because processing is organized around a single-GPU queue.

Pros and cons

Pros

Full Studio quality, three export formats and lyrics, BPM and key detection are included in the free tier.

Credits do not expire, have no daily limit, and failed API jobs are refunded.

Broad audio, video and URL input support covers common creator and developer pipelines.

Deletion within 24 hours and no AI training use reduce retention concerns.

Cons

Three free songs per month is restrictive for frequent users.

Uploads stop at 100 MB and 20 minutes, and TikTok requires a saved video upload.

The single-GPU queue may not provide the throughput that batch teams need.

Users must manage rights for every submitted recording.

Alternatives

AudioShake is the better fit when you need broader platform coverage, a starting allocation of 200 credits, ten-minute files and lyric editing.

LALAL.AI suits users who prefer a freemium service with a relaxed queue and 200 MB per-file uploads.

Krisp Voice Isolation is aimed at early-stage or enterprise deployments requiring speech-isolation models and server-side batch processing.

Altered Real-Time Pro is preferable for desktop speech enhancement and voice skins rather than song-stem separation.

clir Focus is an iOS-only option for users who want session-based voice isolation on a phone.

HitPaw Vocal Remover fits buyers seeking a paid desktop/web license with AI credits and technical support.

VocalDrop is the choice for a free desktop workflow with no track limit, watermark or account requirement.

Demixr is suitable for lighter web use when a 10-minute monthly allowance and queued processing are acceptable.

Verdict

AIVoiceSeparator is a strong choice for creators and developers who value full-quality separation, flexible inputs, optional lyric transcription and credits that never expire. Choose it over alternatives when occasional use, privacy-conscious processing and API access matter most. Look elsewhere if you need native desktop or mobile software, files beyond 20 minutes, or dependable high-volume throughput without arranging dedicated capacity.