报告错误/功能要求
AI Music Generator AI Song Generator AI Music Video Maker AI Beat Maker AI Lyrics Generator AI Jingle Generator AI Karaoke Maker Vocal Remover More →

AI Vocal Isolator

商业用途OK 380+模型 无水印 不需要注册
型号 :
+ GPT-5, Claude, Gemini

拖放音频文件, 或单击浏览

支持MP3、WAV、FLAC、OGG最多可达500MB的MP3、WAV、FLAC、OGG

未选中文件
Upload a song to separate it into individual stems: vocals, drums, bass, and other (instruments). Each stem gets its own player and download button.
~1,000 tokens base + ~50/sec of audio (~3K for a 1-min track)
Separated Stems

你能做什么,你能够做什么

解析 Vocals

将音轨从任何歌曲中分离出来。 适合创建 acapellas、 remixes 或卡拉OK 版本 。

获取乐器

删除声音以获得干净的工具轨迹。 用于背景音乐、 封面或练习会话 。

4 独立的 Stems

分成声响、 鼓声、 贝斯 和其他乐器。 每根干叶都有自己的音频播放器和下载按钮 。

多重格式

MP3, WAV, FLAC, 或 OGG 文件上传至 5000MB 。 使用任何音频格式和质量级别 。

快速处理

我们的GPU的AI动力干干分离运行,大多数轨道在一分钟内处理完毕。

下载每个 Stem

每个分离的干茎都可以单独下载。只要声音,鼓,或所有四根。

如何使用

上传您的音频

拖放音频文件或单击以浏览。 支持 MP3、 WAV、 FLAC 和 OGG 最多可达 5000MB 。

单击删除 Vocals

按下按钮,我们的人工智能将分析音频 并用深层次的学习 将音频分离成个别的发源

每个列列预览

每个干叶都有自己的音频播放器。

下载你的根

下载任何或全部分离的根。 用于混音、 卡拉OK、 音乐制作或练习 。

记脚成本

行动和行动行动的行动使用时代数Output
Stem 分离~50/sec4根:声响、鼓声、低音、低音、其他

Free.ai有400+人工智能工具,

探索全部工具 获取更多当当量
高级选项
结果成果成果成果成果成果成果成果成果成果成果
声调越来越低 Get More Tokens
Want better results? 模型 (GPT-5, Claude, Gemini) deliver higher quality. View Plans

❤️ Love this tool? Share it!

< a href=" "/ signup/" style=" "color:#16A34A" > 签名 以获得查询链接, 并获得每个朋友25,000个象征性的 。

还要吗? 每日5K象征性的免费签名 + 10K奖金
签署自由

处理您的请求...

Isolate vocals from any song with AI. Get the acapella, leave the music. Demucs MIT, browser-only, no upload limits.

如何使用 AI Vocal Isolator

1
输入输入

键入文本、上传文件或描述您想要的东西。不需要账户 。

2
点击生成

我们的人工智能使用最佳的开放源码模型,在秒内处理你的请求。

3
下载共享( S)

下载、复制或分享您的结果。 免费个人和商业使用 。

通过 API 使用此工具

从您自己的代码中自动启用此工具。 OpenAI 兼容的 REST 端点、 Bearer-tok 异常点、 不需要额外的 SDK 。 Token 成本符合网络界面 。

API 文件 获取 API 密钥 
curl -X POST https://api.free.ai/v1/music/generate/ \
  -H "Authorization: Bearer sk-free-..." \
  -H "Content-Type: application/json" \
  -d '{"prompt": "upbeat synthwave instrumental", "duration": 30}'

相关免费 相关 AI 工具

AI Music Generator
AI Song Generator
AI Music Video Maker
AI Beat Maker
AI Lyrics Generator
AI Jingle Generator
AI Karaoke Maker
Vocal Remover

AI Vocal Isolator — FAQ

It pulls the vocal track — singing, speech, or any voice — out of any song while leaving the music behind. The output is a clean acapella WAV. Same Demucs source-separation model that powers Moises and Splitter.ai.

/music/vocal-remover/ keeps the music + drops the vocals (gives you an instrumental / karaoke track). AI Vocal Isolator does the opposite: keeps the vocals + drops the music (gives you an acapella). Same underlying model, opposite stem.

Remix it over a new beat, layer harmonies in a DAW, sample for hip-hop production, transcribe lyrics, train a voice clone, or just listen to the raw vocal performance without backing.

Very clean on dry, center-panned vocals. Some residual reverb tail or bleed when the vocal is drenched in delay or hard-panned. Live recordings are weaker than studio masters because the vocal mic picks up the room sound.

MP3, WAV, M4A, FLAC, OGG, AIFF, and any common audio format. Max 50 MB upload (about a 5-minute 128 kbps MP3).

6-15 seconds per minute of audio on our A100. A 3-minute song completes in roughly 25-40 seconds.

Personal use (study, practice, fan remixes for non-commercial enjoyment) is generally fair-use OK. Distributing the isolated acapella, sampling it commercially, or publishing it on streaming platforms requires a sync or sample license. You're responsible.

~3,000 tokens per song, flat rate (per-request, not per-second). Free daily-pool tokens cover anonymous use; signed-in users get 5K/day.

Not in AI Vocal Isolator — pipe the output to /music/pitch-changer/ to transpose ±12 semitones. Useful for matching a different key when remixing.

Same Demucs model = same quality. Moises is $4-12/month with a track quota. LALAL.AI is per-track. Free.ai is unlimited within the daily token pool, no subscription, no signup required.

Demucs is near-perfect but not perfect. Heavy reverb tails on the vocal sometimes get mistaken for music, and music that's closely tied to the vocal (e.g., orchestral swells under a sung note) can leak. For best results, start with the cleanest studio master you can find.

Yes — POST audio to /v1/audio/separate/ with `mode=vocals`. Returns a job_id you poll, plus the WAV URL when done. Docs at /api/.

免费注册一万个符号

创建自由账户

无需信用卡

你会如何评分这个工具?

Love this tool? Share it!