Compare how sound changes in 39 short recordings. Listen while following the waveform and frequency content.
The atlas covers all 16 synthesis and effect techniques in the current guide and all 12 implemented effect families. These 39 recordings define the current comparison set. Other parameter choices and combinations can generate further studies.
Actual left and right PCM samples over approximately the previous 85 ms, shown at a fixed ±0.25 FS scale. Compare envelopes, periodicity, noise and gating. Values outside the visible range are clipped at the display edges; the numerical peak reports their measured level.
A 4096-point FFT with a Hann window is grouped into 96 logarithmic bands. Bars show each band’s RMS level in dBFS. Compare harmonic spacing and the attenuation of high frequencies by filtering.
The spectrogram shows the previous 4 seconds, with the present at the right edge. Follow reverb tails, FM sidebands and repeated captured components over time.
The 3D view shows the paths of sources and their effect-generated voices. Bus meters in the landscape videos show measured source PCM levels; RMS and PEAK in the portrait videos measure stereo output. Dot size and panning gain are distinct from measured signal amplitude.
On phones, the waveform, spectrum and history are stacked in a portrait video. Audio is identical to the landscape version. The 3D window is taken from the original application recording.
Each recording uses 48 kHz audio and 30 fps video: 1600 samples per frame. Analysis uses actual PCM up to the frame timestamp. The 3D scene uses an illustrative studio layout, and sound is the application’s stereo audition. It does not reproduce venue acoustics or use binaural rendering. Most entries compare a 4-second reference with 4 seconds of processing; examples with several changes show their own timeline.
Waveforms and spectra measure final stereo PCM. Bus meters measure PCM after master gain, before panning and final soft clipping. Video audio is encoded as AAC from that PCM. The zero-rate read-head example holds a constant sample rather than sustaining an audible tone.
Use one periodic component as a reference, then raise it one octave.
Change sine to triangle while retaining pitch and position.
Change sine to sawtooth to compare harmonic density.
Change sine to square; pulse-width modulation is not included.
After a sine reference, move through triangle, saw, square and sine in one-second steps.
Open a sawtooth low-pass in steps from 500 Hz.
Retain a 900 Hz cutoff and raise resonance from 0.05 to 0.8.
Blend 80% seeded noise into the periodic source.
Change the second half of the same figure to a short attack and decay with low sustain.
Compare two gates at one pitch, using a 1.2-second attack and 0.9-second release in the second half.
Add a 5 Hz sine modulation of ±1 semitone to a fixed note.
Modulate a sawtooth filter at 0.75 Hz over ±2 octaves.
Add 2 Hz amplitude modulation with depth 0.8.
Move DSP x-position at 0.5 Hz over ±4 m.
At a fixed right-side position, modulate width around 0.45 by ±0.45 at 0.5 Hz.
Orbit the source in the second half and interpolate two fixed timbral anchors.
Keep the same three-note chord and separate it into left, center and right positions in the second half.
Change FM index from 0 to 4 at ratio 2, retaining position and base pitch.
Change FM index from 0 to 3 at a ratio of 1.4142.
In a 16-partial bank, change amplitude slope from 3 to 0.7.
Change the same figure to 16 partials with inharmonic ratios and individual decay in the second half.
Change a sine source to multiplication by a sine at ratio 1.5.
Raise AM depth from 0 to 0.9 at a ratio of 1.5.
Return a signal delayed by 350 ms at the source position.
Send the same 350 ms delay to a child source displaced 4 m right.
Read 100 ms grains from actual input history at 12 grains per second.
Group 160 ms grains at 72 grains per second into eight cluster buses.
At four seconds, capture each voice’s preceding 350 ms and read it repeatedly.
Loop the preceding 750 ms of mixed history at rate −1 in the second half.
Loop the preceding 750 ms of mixed history at rate 0.5.
Stop the read head at one point in the captured waveform.
Repeat short history slices at a 7 Hz cycle and mix them into the source.
Enable an approximately 31 ms feedback path with coefficient 0.7.
Split low, middle and high bands and place them at three positions.
Extract bands near 1, 2, 3 and 5 times the known note, retaining the residual.
Create two complementary masks from amplitude-envelope differences and position them left and right.
Enable a four-delay network with a decay setting of 2.2 seconds.
In the second half, move between two delay configurations at morph values 0, 0.5 and 1.
Apply different all-pass delays to four paths and distribute them around the source.