NOMN

← 工具

FOAM:气泡声学与泡沫合成

¥11,111 / $69.65
常见问题
What platforms does FOAM run on?
macOS (Apple Silicon with untested support for Intel) as VST3, AU, and Standalone. Windows is in beta as of April 2026 but is working on our test machine with Reaper and as standalone. Our most thorough testing is with Reaper, but we have tested with Bitwig, Plogue Bidule, and a number of other platforms. Please report bugs if you encounter anything strange with other hosts.
Does it support MPE?
Yes. Per-note pitch bend, slide, and pressure map to bubble size, pitch sweep, ring time, emission rate, and amplitude.
Is there a free trial?
Yes. Once you've registered an account, you can get a trial code and start a trial by clicking on the "Unlicensed" message within the Foam application itself.
Any tips to get started?
To hear the range of sounds you can produce with FOAM, a quick way to explore is connect a midi controller of some kind, turn up chaos and bounce around and hear the results. Right-clicking any knob will lock it in position and leave it unaffected by the jumps/modulation of the Chaos knob. Chaos also will change to a more traditional LFO-based modulation source if you right-click on it. We recommend locking the Fluid, Pitch, and Topology parameters as a starting point.
Can you automate all of these knobs in a DAW?
Yes, and this is definitely how you can dial in very particular sound-shapes and harmonies, etc. A first test is draw some long curves with Rate / Size / Ring involved. If you're in harmonic/melodic mode try drawing changes in the Root / Scale for chord changes..
Why aren't there any presets?
Answer are how questions are lost to ideology :)
There is a robust user preset system and a way to make time-based morphs between presets
FOAM is getting stuck or hanging when I get over X bubbles or automate the size / ring / rate knob to max? Is it broken?
We've made the choice to let users throttle their version of FOAM themselves and manage threading a bit since at the moment people are running computers with wildly different CPU capacity. If you're having issues, the remedy is to click the CPU menu and lower the maximum bubble count to something that seems safe given what you're seeing and then slowly raise it. This is also important if you're trying to run multiple instances or layer FOAM into a dense session with many plugins, etc. FOAM uses multiple threads to chew through bubble math in parallel -- depending on your machine you might also find that more or less threads will yield better results. Users on really fast machines may fund that 100k bubbles/sec is not their theoretical max but for now, that's our limit to avoid extreme CPU spikes for most users.
Why does FOAM seem to use so much CPU? Is it broken? Isn't there a better way to generate foamy sounds?!
Physical modeling has a reputation for being CPU intensive compared to other synthesis methods -- and while this is a well-earned reputation, we are also reaching a point where CPUs on personal computers are capable of doing things that felt unfeasible a few years back. FOAM is not going to function similarly to a sample-based Kontakt instrument that sips CPU and triggers samples and that's not a bug, it's just a different use-case with different limits. You can definitely make amazing wet/foamy sounds on a modular synth, using samples and wavetables, etc but to access and move through the range of possible states FOAM can produce as a creative palette- FOAM is a unique instrument for doing this.
Why doesn't FOAM sound realistic enough to replace samples of crashing waves, running water, dripping faucets and stalactites? Champagne?
FOAM produces mathematical representations of a natural phenomena for creative use but definitely will not replace actual foley for diagetic sound design work. That said, it can produce many sounds that are psychologically quite linked to these types of sounds, which is tremendously useful for sound designers thinking about getting a listener into a more imaginary world that is linked closely to the natural world. With automation you can get quite close to a crashing ocean wave, but we don't imagine that the purpose of FOAM is to replace stock libraries or field recordings by any means. It's more about exploring the edges of these sounds than trying to recreate them 1:1.

过去二十年里,真正全新的合成方法几乎没有出现。FM、粒子、波表、击弦和弓弦的物理建模——这些范式在90年代初就已确立。此后,大多数"新"插件不过是对现有理念的改良、组合、更好的界面或索引。

FOAM是一次开辟新领域的尝试。

气泡声学是自然界中声音最丰富的现象之一,也是合成领域中探索最少的。其物理学已有充分文献记录,但作为声源却缺乏深入理论研究,主要因为行为过于混沌,没有人试图对其进行实时建模。

一道退去的浪的声音不是五个或十个声源。而是数千个跨越广泛行为范围的相互作用,其总体构成了声音纹理的丰富性。FOAM直接实现了这种物理。不是采样。不是近似。是当气体遇到液体时发生的一切的真实声学建模——从最初的薄膜破裂到空腔共振再到粘性阻尼。

这是一个复杂的工具,我们建议配合自动化曲线使用以获得最大的细节和形态变化。它也是一个非常有趣的表演合成器,可以用MPE演奏或用模块化装置驱动。对于噪音爱好者,有一个彩蛋等你发现。
--
## 工作原理

每个气泡是一个物理方程,不是一段录音或粒子。

FOAM建模了气泡生命的11个不同声学阶段,从最初的薄膜破裂到共振、坍塌和衰减。每个阶段由其自身的物理行为支配,每个参数都映射到真实的流体力学。

每个参数都输入声学公式。提高粘度,数学就会改变:小气泡在粘稠液体中迅速消亡而大气泡振响更久——与自然界一致,在当前气泡声学研究所知的范围内。

在峰值密度下,数千个合成声部在多个CPU核心上同时渲染,产生的纹理具有采样回放或波表合成无法以类似方式进行创意塑形的时间细节层次。

由于气泡的计算方式,为获得最直接、瞬态最丰富的响应,我们建议缓冲区大小为64-128采样。当CPU大幅提速后,将有可能不受缓冲区限制地规范化行为,但目前这是我们能创建的纹理规模与使FOAM在操作系统内实际可用之间的最佳折中。

--
## 引擎

最多16,384个同时发声,带有级联建模——母气泡触发相互作用的子气泡链。可配置的CPU调控器让你设定声部密度上限——在现代硬件上完全开放,或在密集会话中适当降低。随着密度增加,群体动力学产生集体频率偏移,从离散爆裂声到连续泡沫纹理的过渡自然地从大量独立物理交互中涌现。

范围从单个水滴到致密泡沫筏。从水到蜂蜜到水银。从细腻的雨声到噪音墙。如果你想要,几乎可以做出逼真的河流声。如果你不想,可以进行完全的纹理探索——用幻想尺寸的气泡,也许还能找到一个kick或sub-bass。
--
## 参数
Bubble Physics
- Rate:发射速率与能量。
- Size:基础气泡尺寸(音高)。
- Ring:振铃时间/衰减特性。
- Cascade:子气泡链深度。
- Glide:音符间滑音。

Realism
- Shimmer:表面波纹调制。
- Natural Balance:薄膜破裂与空腔坍塌混合比。
- Jets:二次喷射建模。
- Mirror Depth:浸没深度与表面反射。

Texture
- Stereo Width:立体声展宽。
- Magic:心理声学强度(幻影sub-bass、耳蜗失真、双耳粗糙度)。

Parameter Modulation
- Chaos:参数随机化,支持逐旋钮锁定。
- Drift:缓慢的布朗随机游走。
- Scatter:逐声部随机化。

Liquid Identity
10种液体材质预设:Water、Seawater、Soap Solution、Glycerol、Honey、Olive Oil、Mercury、Ethanol、Molasses、Custom。
- Liquid Mass:液体密度(手动)。
- Tension:薄膜强度(手动)。
- Viscosity:阻尼特性(手动)。

Foam Topology
- Topology:结构泡沫模拟(粗化、坍塌事件)。
- Foam Age:泡沫结构的时间演化。
- Wetness:泡沫中的液体含量。
- Stress:泡沫网络的机械应力。
- Yield:屈服应力阈值。
- Jamming:2D到3D结构转变。
- Burstiness:事件时序统计(规则vs聚集)。
- Aging Rate:结构演化速度。

Surge Modulator
- Surge:时间包络形状(Wave、Pummel、Roil、Gush、Burst)。
- Surge Time:包络持续时间(1–30秒)。
- Surge Depth:调制强度。

Froth Gen
6种发射器模式控制气泡随时间的释放方式:
- Poisson:自然随机发射。
- Nucleation:碳酸化风格释放。
- Foam Raft:雪崩动力学。
- Turbulent:间歇性湍流。
- Melodic:音高稳定的发射,用于旋律演奏。
- Harmonic:音阶量化发射,用于谐波内容。

Pitch Gravity
气泡基于尺寸产生自身的共振频率,但为了创意应用,你可以将其拉向音高集合。
- 76 Scales:来自世界各地的传统,加上微分音系统、异调律调音和合成构造。
- Root:移调至任意调。
- Quantize:音阶符合强度。

Voicing Dynamics
当数千个气泡同时发声时,它们的叠加方式改变了声音的特性。FOAM的混音部分在两个阶段提供控制。
- Gain Scaling:随群体增长的气泡振幅平衡方式。
- Freq Summing:低频和高频气泡是平等贡献还是按物理缩放。
- Spatial:空间位置对声部振幅的影响。
- Transients:初始爆裂攻击在缩放中的保留程度。
- Drive:压缩器前输入增益。
- Multiband Comp:8段并行压缩器(63Hz–8kHz),逐段混合与自适应动态。
--
## 功能与格式
主要功能:
- 完整MPE支持:逐音符弯音、滑动和压力,用于表现力控制气泡尺寸、音高扫描、振铃时间、发射速率和振幅。
- GPU加速可视化:气泡群体、密度和频谱内容的实时显示。
- MIDI Learn:右键任意参数分配CC控制。
- OSC输出:将气泡事件流传输到外部应用,用于同步视觉或数据声化。
- 预设系统:保存和调用配置。预置了统计上具有差异性的音色,而非精心设计的方案。这是供你自行探索的开放领域。

兼容性:
格式:VST3、Audio Unit (AU)、Standalone。

操作系统:macOS(Apple Silicon + Intel)和Windows(2026年4月起实验性支持)。
--
## 关于缓冲区大小

FOAM是一个随机物理模型。每个气泡都由流体力学方程支配的过程产生,每秒有数千个相互作用。你的DAW音频缓冲区大小会影响输出。就像原声乐器一样,FOAM的特性会响应它所处的演奏条件。

这不是bug。这是在实际尺度上实时建模混沌物理系统的本质,与粒子或波表不同。

在较低的缓冲区大小下,引擎解析更精细的时间细节——更多的瞬态定义,更连续的纹理。

为获得最直接的响应,请将缓冲区设置为64或128采样。这可能会使围绕大型采样库构建的会话变得复杂,因为后者受益于较高的缓冲区大小。如果你需要在512或1024下工作,声音会有所不同——对你的情况来说不一定更差,但不同。