Contents

Introduction

A screenshot of Lost Technology.

Figure 1: A screenshot of Lost Technology. Analyzers are on top, Modulators in the center and Synthesizers in the lower corners.

Lost Technology transforms sounds by taking them apart and then putting them together in a different way. The Analyzer components split the input signals into amplitude, frequency and waveform (harmonic content) sub-signals. The Modulator components combine sub-signals (and audio signals) in various ways. The Synthesizer components merge one sub-signal of each type into a new audio signal.

Internal signal paths of Lost Technology.

Figure 2: Internal signal paths. Rounded shapes represent components, arrows represent signals. "Mod" is an abbreviation of "Modulator". Labels next to the arrows show signal type.

Basic terminology

A signal is a function whose domain is a (possibly infinite) time interval. The length of a signal is the length of its domain interval.

A slice is a signal with known, finite length. Lost Technology divides its input signals into thin slices and then constructs output signals out of those slices.

An audio signal is a real-valued signal that represents a sound. Audio signals that contain values greater than 1 or less than -1 are overloading and may not be processed normally.

An amplitude signal is a nonnegative-valued signal that represents the loudness of a sound. The value 0 (0%) represents silence, the value 1 (100%) represents maximum volume. Amplitude signals that contain values greater than 1 are overloading and may not be processed normally.

A frequency signal is a nonnegative-valued signal that represents the pitch of a sound. The value 0 (0%) represents 1Hz, the value 1 (100%) represents the maximum frequency of Lost Technology (22kHz). Frequency signals that contain values greater than 1 are overloading and may not be processed normally.

A waveform signal is a signal whose values are audio slices; it represents the overtones of a sound. Waveform signals whose values contain values greater than 1 or less than -1 are overloading and may not be processed normally.

Strictly speaking, a cycle is just a slice of an audio signal. However, the term is used to denote audio slices that Lost Technology has chosen and manipulated to make useful "building blocks" for sound generation. The concept of the cycle in Lost Technology is similar to the concept of the grain in granular synthesis. (In fact, the stuff Lost Technology does could be described as "granular analysis and synthesis".)

The current value of a signal is the value the signal function has at the current point in time. That is, current_value(signal) = signal(current_time).

Slice operations

This manual often shows signal slices being used as operands in addition, multiplication and other mathematical operations. To understand the meaning of these expressions, it is useful to think of a slice as a list of numbers. Applying an operation to two slices is equivalent to applying the operation to each pair of numbers that are in the same positions in the two lists and making one new list out of the results. For example, if sliceA = (3, 4, 2) and sliceB = (5, 1, -3), then sliceA+sliceB = (3, 4, 2)+(5, 1, -3) = (3+5, 4+1, 2-3) = (8, 5, -1).

The Analyzer

The Analyzer produces an amplitude output signal by measuring the average signal level of its audio input signal. The Analyzer also produces a frequency output and a waveform output. Those two outputs are created through a process that chops the audio input into slices in accordance with certain rules, as will be explained in the following subsections. The frequency output is formed by measuring the length of each chopped-off audio slice and converting it into a frequency signal value. The values of the waveform output signal are the chopped-off slices themselves.

Terminology

The following list contains definitions of some important terms that are specific to the Analyzer.

Algorithm

Signal flow within the Analyzer.

Figure 3: Signal flow within the Analyzer. Parameters are shown in the sub-components they control.

This section describes the Analyzer's sub-components in signal-flow order, that is, each sub-component is described before the sub-components it feeds signals to.

The amplitude tracker approximates the amplitude of the input by setting OA to a weighted average of OA and AIL every time a new sample is sent to the audio input. The formula looks like this: OA := w*OA + (1-w)*AIL. For increased stability, the amplitude tracker can be configured to react quicker to increases in AIL than to decreases. This is done via the parameters AIncLag and ADecLag. The weight w of OA is equal to AIncLag when AIL is greater than OA and to ADecLag when AIL is less than OA.

The frequency and waveform update gate (FWUG) decides when the input is loud enough to extract frequency and waveform information from. While OA is less than GateLvlA and AIL is less than GateLvlS, the FWUG will keep the FWUT (see below) disabled.

The frequency and waveform update trigger (FWUT) decides when it is time to update OF and OW. Its decisions are based on minimum and maximum frequencies set by the user and on detection of cycles in the input signal.

Within the FWUT, a cycle is defined as a slice that begins and ends with a zero-crossing and contains a low valley followed by a high peak. If the input is perodic and reasonably well behaved, the length of the detected cycles will be approximately equal to the period length. Since the base frequency and harmonic content of a periodic signal are determined by the length and form of its periods, detected cycles can be used to extract frequency and waveform information from the input.

The FWUT uses the following cycle-detection rules:

1: If the waveform buffer is full, UPDATE.
2: If elapsed time since the last update is greater than 1/MinF, UPDATE.
3: If elapsed time since the last update is less than 1/MaxF, DON'T UPDATE.
4: If RIL has not been less than LowTrig and then greater than HighTrig at least once since the last update, DON'T UPDATE.
5: If RIL has crossed from positive to negative, UPDATE.
6: Otherwise, DON'T UPDATE.

The frequency tracker approximates the frequency of the input by setting OF to a weighted average of OF and DF every time a frequency and waveform update is made. The formula looks like this: OF := FLag*OF + (1-FLag)*DF.

The waveform tracker approximates the waveform of the input by setting OW to a weighted average of OW and DW every time a frequency and waveform update is made. The formula looks like this: OW := WLag*OW + (1-WLag)*DW.

How the Analyzer interprets its input signal.

Figure 4: How the Analyzer interprets its input signal. The blue curve shows how RIL varies with time. The second update (at T3) occurs at the first positive-to-negative zero-crossing after RIL has been less than LowTrig (at T1) and then greater than HighTrig (at T2) and at least 1/FMax seconds (Tmin - T0) has passed since the first update (at T0). DF is equal to 1/(T3 - T0). DW is the signal slice between T0 and T3. If no other update rule had applied, an update would've been triggered when 1/FMin seconds (Tmax - T0) had passed.

Parameters

Displays

The upper pair of display boxes show the current values of the Analyzers' amplitude and frequency outputs. The displays are labeled "PreAmp" and "PreFrq" because the displayed values are the ones the signals have before they pass through the Modulators (pre-modulation).

The Modulators

A Modulator is a component that takes two input signals and produces one output that is a function of the inputs. Lost Technology contains modulators that operate on amplitude, frequency, waveform and audio signals.

The values of modulator input signals are positive or negative percentages of fixed maximum levels. (See Basic terminology.) Amplitude and frequency modulator signals use the nonnegative interval [0,1]. Waveform and audio signals use the interval [-1,1]. The fact that modulator input values have magnitudes less than or equal to 1 is important. For example, it means that Mult - the modulator function that multiplies its input values - will never produce an output value higher than the highest input (0.5*0.5 = 0.25, 1*1 = 1).

The primary signal of a Modulator is the input signal that comes from an Analyzer or Synthesizer that is on the same side of the plugin as the Modulator. The secondary signal is the signal that comes from the other side.

To make the abbreviations for amplitude and audio Modulators different, the latter are also called output Modulators. The justification for this is that the signals produced by the audio Modulators are sent directly to the plugin outputs.

Parameters

Functions

These functions are designed to avoid internal overloading. A Modulator output signal will not be overloading unless at least one of the input signals was overloading.

In the function expressions in the following list, m represents the ModMix setting, s1 the value of the primary signal and s2 the value of the secondary signal.

The Synthesizer

The Synthesizer produces an audio output signal by generating audio slices and putting them in a buffer. It feeds samples from that buffer to the audio output one by one. When there are no samples left, the Synthesizer retrieves the current values of its amplitude, frequency and waveform inputs and uses them to generate a new audio slice. The Synthesizer then places the new slice in the buffer and continues to feed samples to the output.

Algorithm

Signal flow within the Synthesizer.

Figure 5: Signal flow within the Synthesizer. Parameters are shown in the sub-components they control.

This section describes the Synthesizer's sub-components in signal-flow order, that is, each sub-component is described before the sub-components it feeds signals to.

The amplitude adjuster and frequency adjuster modify the values of the amplitude and frequency inputs by multiplying them by a gain coefficient (given by AGain or FGain), adding an offset (given by AOffset or FOffset) and then limiting the result to the interval [0,1]. The entire adjuster operation can be expressed like this: output := min(1, max(0, gain*input + offset)).

The audio renderer is activated when the Synthesizer needs more samples to send to the output. The renderer gets amplitude and frequency values from the adjusters and then generates an audio slice by sampling from the waveform input. The number of samples taken from the waveform input is determined by the frequency value (together with the current output sample rate and Oversmpl setting). The signal level of the rendered slice is determined by the amplitude value.

The output smoother reduces discontinuity at the points in the output signal where two consecutive rendered slices are joined. Discontinuity occurs when a slice doesn't end at the signal level and slope where the next slice starts. This causes a "gap" and/or "kink" in the output and may be heard as a click or crackle. To remove discontinuity, the output smoother modifies samples in the smoothing windows - small sub-slices centered on the points where two rendered slices meet. The size of the smoothing windows is determined by the SmooWin parameter setting.

Parameters

Displays

The lower pair of display boxes show the amplitude and frequency values most recently used by the Synthesizer. From LT 0.2.5 and onwards, these values include offset and gain applied by the Synthesizer. The displays are labeled "PostAmp" and "PostFrq" because the displayed values are the ones the signals have after they have passed through the Modulators (post-modulation).

Using Lost Technology

Connecting with the host

Lost Technology can be used in a number of different IO configurations. It can process two distinct mono sounds and output either two different mono sounds (2xMono->2xMono) or one composite stereo sound (2xMono->Stereo). It can also take a single stereo sound and turn it into either two distinct mono sounds (Stereo->2xMono) or a different stereo sound (Stereo->Stereo). Single mono input and/or output is also possible.

If your host application doesn't support 2xMono plugin input, but has a reasonably sophisticated mixer, you can emulate 2xMono with panning. Pan the first input track all the way to the left and the second input track all the way to the right, then send both input tracks to the same stereo track and connect the plugin to that track.

When given a single mono input, Lost Technology will send the same input signal to both Analyzers. When producing a single mono output, Lost Technology will output the signal from the first (left) output Modulator.

Setting parameters

Click the small square button above and to the left of a parameter control to link that control to the equivalent control on the other processing channel. When one of the controls in a linked pair is moved, the other one will move to the same value. Controls can also be linked component-wide, using the buttons in the upper right corners of components, and globally, using the buttons in the upper (pre-modulation) display boxes.

To set buffer sizes, click the box labeled "buffer size" in the upper left corner and choose a size from the popup menu. The display boxes to the right of the selection box show memory use, minimum Analyzer output frequency and maximum bufferable waveform length for the current buffer size setting.

When the "help" button in the lower right corner is selected (solid white), a brief description of the most recently modified parameter / selected modulator function is displayed to the left of the button.

Creating sounds

The algorithms Lost Technology uses to process sound are based on very easy math and are therefore not particularly intelligent. If the input is too complicated for the algorithms to deal with, the output will be too garbled and atonal to fit as a melodic part in a song. Chaotic output can be made somewhat more coherent through careful parameter tweaking, but in most cases the improvement will not be large enough to make the output work musically.

If you intend to use Lost Technology to produce any kind of melodic or harmonic sound, the best strategy will usually be to start with very simple inputs and then gradually increase complexity and tweak parameters until desirable results are achieved. There are a number of simple/complex dualities that can be used as guidelines for the sound creation process:

There is an additional guideline that applies when you use two distinct input sounds, one of them being the primary track that you want to modify and the other a modifier signal. If you are only going to use some parts of the modifier, like the frequency but not the amplitude or waveform, you should give the other parts simple fixed values. In the frequency-only case, this means that the modifier should be a pure note that plays continuously at constant amplitude, even when no key or other input device is pressed or otherwise activated. Hitting a key or performing some other input operation should only change the pitch of the sustained modifier note. The point here is that when the modifier signal is silent, no frequency can be extracted. If that is the case when a note is hit in the primary track, then the primary will not be modified as intended.

Secret techniques

This section lists some interesting but possibly non-obvious ways to use Lost Technology.

There are a couple of different ways to make the plugin "hold" frequencies and waveforms. If GateLvlA and GateLvlS are both turned up high, then the FWUT will be disabled most of the time and the OF and OW will not be updated. If FLag and/or WLag are turned up all the way, then the detected frequency and/or waveform will have weight 0 in the tracker calculations. These hold methods have similar but subtly different effects; for example, the first method allows F&W updates to be turned on and off by setting the input amplitude high or low.

Turning up the FLag a good bit (like 90-something percent), but not all the way, can produce a portamento effect.

The Comp modulator function lets Lost Technology act as a waveshaper. To use it as such, send the signal you want to shape to Input 1 (left) and send a shaping signal to Input 2 (right). Set waveform Modulator 1 to "Comp, 100%" and output Modulator 2 to "Add, 100%". The other Modulators should be left at "Add, 0%". The waveform of the signal sent to Input 2 will now work as a morphable waveshaper function.

If AOffset is set to a positive value, Lost Technology will continue to produce sound even when the inputs are silent. A similar effect can be achieved by turning up ADecLag all the way, so that the Analyzer's output amplitude cannot decrease.

Turning up the AGain is not equivalent to putting an amplifier/dist after Lost Technology in the effect chain. The signal that is amplified and possibly clipped by the amplitude adjuster is the amplitude input signal, not the Synthesizer's output signal. A high AGain setting will drive the Synthesizer toward output with constantly loud, "flattened" dynamics but will not lead to clipping in the audio output.

The FOffset and FGain controls let Lost Technology act as a crude pitchshifter. Oversampling usually produces clearer results, especially when shifting to a higher pitch.

Set the amplitude and waveform modulators for Input 1 to "Add, 100%". Set the output modulator for Input 2 to "Add, 100%" as well. Leave the Input 1 frequency modulator at "Add, 0%". Send a looping, sustained-waveform (e.g. pure sine) synth pattern to Input 1. Send whatever instrument you want to sex up to Input 2 and enjoy the chiptune arpeggiator riot. Try "Fine, 100%" on the frequency modulator for a different (octave-jumping) arpeggiator effect.

Project history

Compatibility, license and contact info

Lost Technology should work fine on PCs with non-ancient processors (Pentium II generation and forward) and a recent NT-based version of Windows (2000 Pro, XP or Vista). It is released under the MIT/Expat/X11 License (see below).

The plugin author maintains a web page from which the latest version of the plugin can be downloaded. Emails to the author can be sent to rabiteman_2000@yahoo.com. Please begin the subject line with the words "Lost Technology".

License

Copyright © 2006-2007 Johan Sarge

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.