New! Feedback Killer AI for Android phones and tablets is on sale, and on iPhone / iPad too. Free 7-day trial
Feedback Killer AI 的界面
AI 开启 AUTO NOISE 工作中：正在切除 2.5 kHz 的啸叫
  • Windows
  • macOS
  • VST3
  • AU

Feedback Killer AI

面向人声和讲话麦克风的 AI 啸叫抑制，一键操作（VST3 / AU）

฿2,199THB≈ $65（估算，以泰铢收费）

每个密钥 2 台电脑
永久有效
已发放密钥 6

使用优惠码 TSM10 仅需 ฿1,979

前往购物车

免费试用 7 天

全部功能，无需注册。购买后在同一安装中输入密钥即可。

安装方法（英文）

付款

  • 信用卡 / 借记卡 全球可用

    Visa · Mastercard · JCB · AMEX，通过 Stripe 支付。付款成功后会自动发放许可证密钥。

  • PromptPay 二维码 泰国银行账户

    用泰国银行 App 扫描二维码，并在订单页面上传转账凭证；核对凭证后发放密钥。

价格以泰铢（THB）计。≈ 金额仅供参考，实际由发卡机构换算。 付款相关问题

购买前有疑问？ 联系我们

详情

本产品说明由英文机器翻译而来，请以英文版为准。

一个按钮：打开它，剩下的交给 AI。它会切除啸叫，让陷波避开人声，并去掉多余的低频，且不增加任何延迟。

  • AI 能区分啸叫和人声：啸叫立即切除，持续的演唱长音（即使加了自动修音）保持不变
  • AUTO NOISE：没人唱歌时，压低串入麦克风的乐器声、风扇、空调、交流哼声和底噪；有人唱歌时几乎不动声音（0.5 dB 以内），没有机器人般的失真
  • 主唱、和声和讲话，每一种人声都完整保留
  • 零延迟：主扩和返听混音都可放心使用
  • 1/24 倍频程 RTA、输入/输出电平表，并显示正在切除的频率
  • 适用于人声和讲话麦克风（乐器请使用标准版 Feedback Killer）
  • 免费试用全部功能 7 天；一个密钥可用于 2 台电脑

规格

类型VST3 / AU 插件
支持系统Windows · macOS
格式VST3 · AU
最新版本1.2.5
许可证每个密钥 2 台电脑 · 永久有效
免费试用7 天，全部功能，无需注册
系统要求VST3：Windows 10/11（64 位）和 macOS AU：macOS（Logic Pro、GarageBand、MainStage） macOS 10.13 或更高版本
使用手册含截图 · 英文 / 泰文 · PDF

截图

版本历史（英文）

  1. 1.2.5 - VOICE ONLY no longer loses the first syllable of a phrase when a band or crowd is in the mic; a quiet room still closes fully between phrases.
  2. 1.2.4 - Catches howls faster in every band, especially the treble (1-20 kHz), before they get loud, and holds a caught howl while the room is still over the edge (no more howl-cut-howl); strong 200-250 Hz howls are held; new VOICE ONLY switch; the display says BYPASSED while bypassed.
显示全部 (12)
  1. 1.2.3 - Fixed a rare crash when closing the host/program while an update check was waiting for the internet. Sound and everything else are exactly as in 1.2.2.
  2. 1.2.2 - Treble howls (2 kHz and up, above all 4 kHz and up) are caught sooner, even while the voice is louder than the howl: a howl that is building is cut while it is still quiet instead of once it is the loudest thing in the room. No extra cuts on singing or music; AUTO NOISE unchanged
  3. 1.2.1 - Fixed treble howls still leaking through: when a treble howl jumps to the next frequency beside its cut (about 100 Hz away on bright PA systems) the AI cuts the new one at once instead of waiting for it to grow, and a treble howl it already cut is remembered and cut again at once if it returns after the cut lets go. No extra cuts on singing or music; AUTO NOISE unchanged
  4. 1.2.0 - AUTO NOISE runs a new AI that makes every decision itself: it tells singing and speech from the music and noise bleeding into the mic far more accurately, takes the bleed down further when nobody is singing, leaves the voice alone and opens and closes more smoothly. AUTO NOISE is still a single on/off switch (with it on, the plug-in uses a little more CPU) / Help improve the AI (if you turned it on): each clip is now 45 s before to 25 s after the AI's cut, with the technical settings at the cut (nothing personal), still at most 20 a day (details at tsoundman.com/privacy) / Move computers yourself: "Deactivate this computer" in the License window gives the key back so you can activate the new computer at once; a moved key stops the old computer within minutes / Fixed: the window could freeze on a computer left running for days
  5. 1.1.0 - Help improve the AI (only if you say yes): each time the AI cuts a howl, or cuts by mistake, the plug-in sends that mic's sound from 30 s before to 5 s after to T Sound Man to train the AI. Asked once; change it with the HELP THE AI button. No name or licence is sent (details at tsoundman.com/privacy). Sound and features are unchanged
  6. 1.0.4 - Fix: after every instance of this plug-in was removed, adding a new one and opening its window could close the host (DAW / SuperRack / Signal Rack). It came with the online-update version; please update every computer. Sound and features are unchanged
  7. 1.0.3 - Online updates: when T Sound Man releases a new version, the plug-in offers it when its window opens and installs it in one click (it downloads and checks the installer itself). Mac installers are signed with an Apple Developer ID and notarized by Apple, so they open without a warning. Sound and features are unchanged
  8. 1.0.2 - About a third less CPU with NOISE off (the noise-reduction filters run only the part that keeps learning the room); the sound is unchanged sample for sample. Switching NOISE on warms the filters for 0.15 s before the fade, without a click. Steadier activation: at most 4.5 s per try to reach the licence server, closing during an activation never hangs, and every legacy key works
  9. 1.0.1 - Easier activation: a Load key file button (pick a file with the key, no typing or pasting) and a Paste button in the License window, stray characters around the key are cleaned off, activation messages always on top, and the licence server reached through tsoundman.com first, which fixes Could not connect to the licence server on some networks
  10. 1.0.0 - First release: AI howl suppression that leaves the voice alone, and AUTO NOISE

iPhone / iPad · App Store

免费下载，试用 7 天后在 App 内一次性购买（无订阅）。支持 USB 音频接口。

也可以在本网站购买完整版（1 台设备：Android 或 iPhone / iPad），然后在 App 的 Settings > FULL VERSION > Sign in 用 tsoundman.com 账户登录。

Android · .apk 文件

安装方法（图片） ↗

免费下载，试用 7 天，通过 .apk 文件安装。支持 USB 音频接口或设备自带麦克风。