1. The compositional idea
A library of recorded sounds is often treated as a collection of objects to be processed. TimeLine Composer treats it instead as a collection of possible musical events. An event has a duration, an entrance, an ending, a level and a spatial position. By composing these relationships, the same recordings can become an expanding texture, a contraction into fragments, a multilayered dialogue or a form organized around a decisive silence.
The tool does not synthesize a replacement for each source and does not time-stretch the recordings to make them fit. Each scheduled event uses a complete source recording, with brief anti-click fades, gain adjustment and spatial placement. The timbral and gestural identity of the source remains intact; the larger musical form emerges from its context.
2. Composition strategies
Choose a Composition preset as a musical proposition, not as an effect. Each preset coordinates source-duration choices, density, overlap, silence, conductor movement, spatial behavior and moderate dynamic shaping. You can inspect and alter its values in Edit_composition_controls.
| Composition preset | Formal idea | Characteristic behavior |
|---|---|---|
| Emergence | From sparse, delicate entrances toward a wider sound field. | Short-to-long durations, increasing density, crescendo and expanding space; early breathing silences. |
| Accumulation | Polyphonic growth through staged entrances. | Independent layers thicken toward the end without planned absolute silence. |
| Breathing Form | Repeated expansion and withdrawal. | Duration and conductor waves, recurring silences and spatial breathing. |
| Climax & Rupture | A late peak interrupted by a grand pause. | Arch-shaped duration preference, buildup, rupture and brief aftermath. |
| Spatial Counterpoint | Several independent voices interact in space. | Layer-specific temporal trajectories, pacing and spatial regions. |
| Fragmentation | A sustained mass breaks into fragments. | Long-to-short durations, decrescendo and increasingly articulated silence. |
| Controlled Chaos | Irregularity within a deliberate overall form. | Loose ordering, strong overlaps, stepped conductor and discontinuous silences. |
| Gesture Crescendo | One short gesture grows toward its climax. | Typically 5–15 s: short-to-long sources, increasing musical energy, density and spatial width; no planned silence. |
| Gesture Decrescendo | One intense gesture relaxes and dissipates. | Typically 5–15 s: long-to-short preference, declining source energy and activity, contracting space; no planned silence. |
| Custom | Build your own formal scenario. | Manually specify temporal, dynamic, spatial, overlap and silence controls. |
Compositional Intensity: from restrained to extreme
Compositional_intensity_(%) changes how emphatically a preset is realized. At 50%, the designed values are used. Lower or higher intensity primarily adjusts density, overlap, ordering and conductor influence; the dynamic settings change only gently. It does not change the requested duration, remove planned silences or alter Gesture_awareness. In Custom, intensity is not applied.
Edit_composition_controls to review and change preset values after intensity scaling. Allow_overlap_within_layers is enabled by default in every preset; disable it if you want one complete recording at a time per layer. Gesture_awareness is in Edit_details.One folder can still create a polyphonic form
All nine presets accept a single source folder. Accumulation and Spatial Counterpoint can turn one folder into three compositional streams. With multiple folders in Independent Layers mode, each folder normally forms its own layer. A layer can be polyphonic, or monophonic if you disable internal overlap.
3. The dimensions of form
These dimensions interact without being equivalent. A bright recording need not be loud; a crescendo need not contain more entrances; a layer can be monophonic while the total composition remains polyphonic. Source selection and event gains are separate decisions. Open Edit composition controls, Edit details, Edit layers and Edit spatial layout to examine them.
4. Reading a possible timeline
Consider a three-layer piece with a late climax. Short isolated sounds gradually give way to overlapping recordings. A structural silence interrupts the accumulation; the final section becomes spatially broad before receding. The example below illustrates a compositional possibility, not an actual render or a screenshot of the script.
Blocks show entrances and durations, not waveforms. The conductor curve is a planning influence on event gain, density and spatial width; the actually loudest passage can occur elsewhere depending on the source recordings.
5. Duration as structure
The Temporal preset biases the selection of recordings by their original length. The duration trajectory is a property of the whole composition, not a command to stretch or shorten the recordings.
| Temporal preset | Compositional motion | What to listen for |
|---|---|---|
| Short to long | Progressively favours longer recordings. | Expansion of gesture, increasing persistence, a possible shift from fragments to sustained form. |
| Long to short | Progressively favours shorter recordings. | Contraction, articulation and fragmentation; with greater density, a possible sense of acceleration. |
| Arch | Short → long → short. | Opening, expansion and return, with the turning point related to the climax position. |
| Inverted arch | Long → short → long. | Compression into a central field of shorter gestures, followed by renewed expansion. |
| Waves | Repeated shifts between short and long recordings. | Several cycles of temporal breathing rather than a single directional development. |
| Controlled random | Probabilistic selection around a preferred duration range. | Open form and material contrast without a prescribed global direction. |
Event Density establishes an entrance-rate tendency; Overlap regulates how much events are superposed relative to their duration. The Conductor Engine can reshape both during the piece. Duration Order Strictness adjusts how closely source selection follows the chosen trajectory, allowing a continuum between tightly directed and more varied material choices.
Nominal duration versus sound duration: the requested time defines the planning horizon. In the final phrase, a source starting before that horizon may continue to its own natural end. This policy does not relax strict internal deadlines protecting planned absolute silences. See natural endings.
6. Natural endings
Target_duration_(s) defines how long TimeLine Composer plans the form and permits new entrances; it is not a hard audio cutoff. Recordings already started in the final phrase are allowed to play through their complete durations. A requested 180-second composition may therefore produce a Sound or Arrangement whose last event ends at 184 seconds.
No new event begins at or after the nominal end, but already started sounds may continue. An optional explicit ending silence begins after the final natural tail.
The automatic Final_fade is Off (0 seconds) by default. A global fade is possible only if explicitly requested in Edit_details → Render and files. Short per-event anti-click fades are retained.
7. Silence as a compositional event
TimeLine Composer distinguishes no new entrance from nothing sounding. A pause in event entrances may still contain long decays or overlapping recordings. An absolute silence is different: it is a planned interval during which no layer may sound. The planner respects those boundaries when selecting whole files, so events must fit before the next silence.
Inter-onset gap
The space between beginnings. A longer event can continue across this gap, maintaining a continuous texture while new entrances become rare.
Absolute silence
A genuine break in the sounding material. It can articulate sections, prepare a climax, interrupt an accumulation, or give an ending its final weight.
| Silence mode | Musical behavior |
|---|---|
| Continuous | No absolute silence is deliberately scheduled; incidental gaps can still occur if the source material is insufficient. |
| Breathing | Numerous relatively short interruptions that suggest phrasing and respiration. |
| Structural | Fewer, longer intervals that can divide the composition into distinct sections. |
| Dramatic | A prominent interruption associated with the climax, creating an expectation or rupture. |
Silence Amount sets the requested proportion of silence in the overall duration. Min Silence, Max Silence and Silence Placement offer finer control, including placement before or after a climax. Ending Silence specifies an intentionally quiet tail at the end of the composition.
The engine also looks for unplanned holes inside intended sounding phrases and tries to fill them with suitable whole recordings. If the available material cannot fit the gaps, those gaps are reported rather than hidden. This is particularly important when the source library is small or repetitions are tightly limited.
When using Non-overlapping Layers, missing entrances can lead to more incidental gaps. The no-overlap constraint takes precedence over repairing these gaps, and any unrepaired silence is reported. An explicit ending silence is placed after the latest natural source tail.
8. Conductor and dynamic range
The conductor curve shapes the composition of the timeline, rather than simply applying a fade to a completed mix. It can affect the level of each entrance, the rate at which events appear and the breadth of their spatial distribution. The resulting sense of intensity may come from gain, simultaneous activity, or both.
How quiet is the beginning, and how strong is the climax?
The user can set the conductor-related dynamic contrast explicitly. In the main dialog enable Edit_composition_controls. In the ensuing composition controls dialog, set Conductor_preset, Climax_position and Dynamic_range_(dB) together. A separate Dynamics_influence control remains available in Edit_details.
The conductor ranges from 0 to 1. Its event-gain contribution ranges from 0 dB at the strongest point to a negative attenuation at the quietest point. For example, 8 dB of Dynamic Range with 0.5 Dynamics Influence means at most 4 dB of conductor attenuation. This is not a compressor and does not guarantee a measured loudness range.
Illustrative curve at 8 dB range and 0.5 Dynamics Influence: maximum conductor attenuation 4 dB. Actual mix loudness depends additionally on the source recordings and overlap.
Compositional Intensity now changes the preset dynamics only gently: each of Dynamic Range and Dynamics Influence is scaled by approximately 0.9–1.1 across the intensity extremes. At 50%, the nine presets use modest conductor attenuation, from 2.7 to 5.5 dB. Larger contrasts remain available by editing the controls directly.
9. Musical Energy Profiling
Recorded loudness and performed musical energy are not identical. A resonant quiet piano note, a bright delicate phrase and a forceful low-register cluster may have very different perceptual identities. TimeLine Composer therefore calculates acoustic descriptors and uses them as soft preferences in source selection, without assuming that any descriptor alone defines musical intensity.
| Measure | Role |
|---|---|
| Loudest 0.5-second level | Captures locally strong material without allowing a long decay to dominate the average. |
| Spectral brightness | Spectral centre of gravity; relatively insensitive to constant recording gain but sensitive to register and timbre. |
| Percussiveness | Compares high-frequency energy immediately around a strong onset with the following 250 ms. |
| Crest factor | Diagnostic only; excluded from the combined score. |
Features are ranked within each source folder. An experimental weighted score (0.4 level + 0.3 brightness + 0.3 percussiveness) provides a ranking; conflicting features can be flagged as mixed. These scores are not objective labels such as soft, aggressive or percussive.
Profile source energy
Select Output → Profile source energy to measure a corpus without composing. The script provides a TimeLine_energy_profile Table, a ranked report, a visualization and an EnergyAudition Sound for each folder (three low-, middle- and high-ranked excerpts). This is useful for checking the ranking by listening before it guides a composition.
Gesture Awareness in normal composition
In Edit_details → Source selection, set Gesture_awareness between 0 and 1. At 0, energy has no influence on source selection. Higher values gently favor source-energy ranks nearer the conductor curve at each event's onset, but duration preferences, phrase deadlines, repetitions and source availability remain primary constraints. The default differs among presets; intensity does not modify it. High values do not necessarily make the music more compelling.
10. Libraries, layers and loudness
One corpus or several interacting voices?
Combined Pool
The selected folders contribute to a shared source collection. The selection process follows one overall temporal trajectory, while each event still retains its source-folder identity for loudness and placement.
Independent Layers
Each source folder becomes a compositional voice with its own duration preference, density factor, manual gain and resting tendency. All voices share the larger conductor and absolute-silence structure.
In Spatial Counterpoint, the layers may adopt contrasting trajectories: one favors short gestures, another longer resonances, and another recurring expansions and contractions. Their spatial positions are adjusted separately in Edit spatial layout, not in Edit_layers.
Library Loudness Balance
Source folders may have substantially different recording levels. The script estimates active-signal level from representative sections, takes a robust reference per library and optionally applies bounded gain compensation. This gives quieter sources a better chance to participate in the form without forcing every file to the same level.
| Balance mode | Behavior |
|---|---|
| Off | Original levels, with neither automatic balance nor manual library gain. |
| Gentle (default) | Half the difference toward a common reference, capped at ±6 dB; manual gain can be added. |
| Full | A stronger correction toward the common reference, bounded by Max_correction. |
| Manual | Only the Gain_dB set in Edit_layers is applied. |
Optional Per-file Leveling is a separate advanced control and is off by default. Preserving the differing dynamic identities of the original recordings remains an important compositional principle.
11. Non-overlapping layers
The option Allow_overlap_within_layers is found in Edit composition controls, directly below Overlap. It is On by default for every preset and Custom, preserving the historical polyphonic behavior.
Yes: Polyphonic layers (default)
Several complete source recordings from the same layer may sound simultaneously. Density and Overlap determine the tendency toward superposition; Ableton may need several Audio Tracks for one compositional layer.
No: One source at a time
Each layer becomes a monophonic sequence of complete recordings. Sources may touch or be separated by gaps, but they may not overlap within that layer. Different layers remain free to overlap one another.
The same three layers can play simultaneously, but inside each layer only one original recording sounds at a time. No recorded sound is cut or stretched to satisfy the restriction.
How to use it: select three folders, enable Edit_composition_controls, choose Organization → Independent layers (one layer per folder), and untick Allow_overlap_within_layers. The existing Overlap slider may still affect spacing, but it cannot create forbidden overlaps.
Multitrack just to get three editable Ableton tracks. Use Send Arrangement to Ableton Live and disable same-layer overlaps. In Combined Pool, multiple folders form one compositional layer, so one track per folder is not implied.If no complete recording fits a free interval, the scheduler delays or omits an entrance rather than cutting a source or permitting overlap. The report lists maximum per-layer polyphony, overlap violations, delayed entrances and any unfilled gaps. In non-overlapping mode each layer's maximum polyphony must be 1.
12. Spatial placement and rendering
TimeLine Composer now treats where a source belongs and how the composition moves through space as different decisions. This distinction matters when a folder represents an instrument or an instrumental family. A percussion group can remain on the left, for instance, even while the composition follows an expanding-space strategy.
Placement: where is the instrument?
Under Edit_spatial_layout, set a Position and Spread for every source region: −1 = left, 0 = center, +1 = right. Spread sets how far event positions may diverge from that center.
Behavior: how does space develop?
In Edit_composition_controls, choose Spatial_preset, such as Random, Left to Right, Expanding, Alternating, Layer-based or Breathing. This defines the spatial trajectory of events through the composition.
The colored areas represent possible event-placement regions. These are not permanent audio tracks: each newly scheduled event receives its own spatial position inside, or partly beyond, its assigned region according to the chosen movement setting.
The essential control: Spatial movement (0–1)
The Spatial_movement setting appears in Edit spatial layout. It determines how strongly the full-field spatial behavior can override the regions you assign.
| Value | Effect | Compositional use |
|---|---|---|
| 0 | The behavior operates inside each group's Position and Spread. | Keep percussion on the left and winds in the center while retaining local motion. |
| 0.5 | Blend the group region with the wider spatial behavior. | Allow controlled excursions without losing the sense of instrumental placement. |
| 1 (default) | The full-field behavior sets the position; manual group positions have no effect except in Layer-based mode. | Free, whole-field spatial gestures independent of source identity. |
Spatial_movement. With values of 1 and a non-Layer-based spatial preset, the planned behavior takes over the entire stereo field. Choose 0 for clear instrumental placement or an intermediate value for partial movement. Layer-based always keeps events region-based, regardless of Spatial movement.How to position separate instruments
- Organize your source material. For instance, use one folder for percussion, one for woodwinds and one for electronic sounds.
- Select a composition preset. Spatial Counterpoint is a strong starting point; other presets can also respect regions once Spatial movement is reduced.
- Enable
Edit_spatial_layout. After the script knows the folders and layers, it opens a separate spatial-layout dialog listing each region. - Set each group's Position and Spread. Example: percussion −0.8 / 0.15, woodwinds 0 / 0.2, electronics +0.8 / 0.15. Set Spatial movement to 0 for region-limited behavior.
- Render and listen. Try changing only Spatial movement to compare a stable instrumental stage with a freer moving field.
One folder, several spatial voices
With one folder, Accumulation and Spatial Counterpoint create three streams. Under Edit_spatial_layout you can edit the three streams independently and choose Shared pool (each stream can select from all files) or Split the files (each source file belongs to one stream). This gives streams more stable material identities, but it does not automatically recognize instrument types: assign instruments to separate folders if their identity must determine placement.
What the stereo or multichannel output represents
Each event has a calculated position; the script does not continuously pan an individual recording over its lifetime. Movement is primarily perceived as successive events occupying new positions. Stereo uses equal-power panning and retains a narrowed stereo image for stereo sources. In four- and eight-channel output, positions map linearly across adjacent channel indices. This is channel-based distribution, not a defined physical surround-speaker layout or Ambisonics.
Multitrack: one mono channel per compositional layer
The fourth Channels setting, Multitrack (one mono channel per layer), creates one multichannel Praat Sound with exactly one mono channel per actual compositional layer. With three Independent Layers, Channel 1 contains all Piano events, Channel 2 all Violin events and Channel 3 all Percussion events. Overlapping events inside a layer are mixed together within its single channel. This is not the same as three editable Ableton Arrangement tracks.
Multitrack bypasses spatial panning, but retains the planned position metadata. Mono sources remain mono; stereo or multichannel sources are converted using (channel 1 + channel 2) / 2. One global peak-normalization factor applies to all output channels. Layer Stems are redundant and are not additionally generated in this mode.
Channels affects Praat audio rendering, not the number of tracks created by the separate Ableton Arrangement importer. In Combined Pool, the multitrack output contains one layer channel even if several folders were selected.13. Creative starting points
These are suggestions for listening and experimentation, not guarantees of a particular sonic result. The source library is always a co-author of the form.
From fragments to a sound mass
Use Short to long, a Crescendo or Late climax, moderate-to-high Overlap, and Expanding Space. Listen for longer events becoming more persistent and layered as the form approaches its climax.
Breathing architecture
Begin with Breathing Form, or combine Arch / Waves with Breathing / Structural silence. Keep density relatively restrained. Let recurring interruptions establish a sense of phrase, punctuation and proportion.
Silence before a rupture
Start from Climax & Rupture, or manually combine Late climax with Dramatic silence. Try Before climax placement in the detailed settings. The silence can transform the subsequent entrance into a formal event.
Counterpoint of recording families
Start with Spatial Counterpoint and use multiple folders in Independent Layers. Give them different duration trajectories and density factors, and edit their regions in Edit spatial layout. Allow the common conductor to create convergence without erasing their independence.
Short, connected gestures
Try Gesture Crescendo or Gesture Decrescendo with a 10-second target. Both avoid planned internal silence, use moderate Gesture Awareness and aim to create growth or release through source choice and texture. Their sound may extend beyond the nominal target to complete natural tails.
Three editable instrumental lines
Choose Piano, Violin and Percussion folders; set Independent layers and turn off Allow_overlap_within_layers. Use Send Arrangement to Ableton Live to receive separate source clips on one Audio Track per layer without splitting overlaps into additional voices.
14. Quick start
- Prepare one to five source folders. The nonrecursive scan reads Praat-compatible recordings in place; it never copies or alters them.
- Choose folders. Leave
Folder_paths_(optional;_separated_by_;)blank to browse, or provide full paths separated by semicolons. Paths determine the actual folder count. - Choose a compositional form. Select one of nine presets or Custom. Enter nominal
Target_duration_(s),Compositional_intensity_(%)andRandom_seed. - Choose overlap policy if needed. Enable
Edit_composition_controls. In Independent layers, untickAllow_overlap_within_layersfor one complete source at a time per layer. Leave it ticked for historical polyphonic behavior. - Optionally refine source energy and space. Under
Edit_detailssetGesture_awareness. UnderEdit_spatial_layout, place the folders in regions or permit whole-field spatial movement. - Choose the destination. Use
Create in Praatfor finished audio,Profile source energyfor analysis,Export Arrangement JSONfor a file, orSend Arrangement to Ableton Livefor background transfer to the Max for Live device. The latter two do not render audio. - Check the report. Inspect nominal versus actual ending, layer polyphony, unplanned gaps and, for Live exports, the expected number of importer tracks.
Allow_overlap_within_layers and select Send Arrangement to Ableton Live. Keep Channels = Stereo; this setting does not control the number of Ableton tracks. See the exact workflow.15. Parameters and defaults
Main dialog: choose a strategy first
| Control | Default | Purpose |
|---|---|---|
Folders_to_browse_(1-5) | 1 | Number of folder chooser dialogs, used only when Folder paths is empty. |
Folder_paths_(optional;_separated_by_;) | Empty | Optional direct folder paths separated by ;; these override the browse count. |
Composition_preset | Emergence | One of nine coordinated strategies, or Custom. |
Target_duration_(s) | 180 s | Nominal planning duration; natural source tails may extend the actual output. |
Compositional_intensity_(%) | 50% | Scales how strongly the selected non-Custom strategy unfolds. |
Random_seed | 1 | Reproducible selections and event placement under unchanged conditions. |
Edit_composition_controls | Off | Open the preset-derived form, including conductor, climax and Dynamic Range. |
Library_balance | Gentle | Compensate for systematic differences in source-library levels. |
Output | Create in Praat | Seven modes: render, plan, export audio, replay a saved plan, profile energy, export Arrangement JSON or send directly to Ableton Live. |
Channels | Stereo | Stereo (default), 4, 8, or one mono channel per compositional layer. This setting does not control Ableton track count. |
Layer_stems | Off | Request synchronized stems in addition to the mix. |
Edit_details | Off | Advanced selection, influences, silence, loudness, render and file settings. |
Edit_layers | Off | Per-library duration selection, density, gain, rest and seed offset. |
Edit_spatial_layout | Off | Positions, spreads, spatial movement and stream-source assignment. |
Draw_visualization / Play_result | On / On | Draw the timeline and play the result automatically. |
Edit composition controls: the musical scenario
When this optional dialog is enabled, it is prefilled with the values generated by the chosen preset and Compositional Intensity. Custom opens it automatically. The values depend on the preset, so there is no single universal default for this dialog.
| Control | Compositional effect |
|---|---|
Organization | Combined Pool or Independent Layers. |
Temporal_preset | Short to Long, Long to Short, Arch, Inverted Arch, Waves or Controlled Random. |
Event_density / Overlap | Frequency of entrances and their amount of simultaneous sounding. |
Allow_overlap_within_layers | Yes (default): polyphonic layers. No: one whole recording at a time within each layer; cross-layer overlap remains permitted. |
Silence_mode / Silence_amount | Type and amount of genuine silence. |
Conductor_preset | The form of the energy / density / spatial trajectory. |
Climax_position | Targeted formal turning point (some curves interpret it differently). |
Dynamic_range_(dB) | Potential conductor-related gain difference between low and high points; affected by Dynamics Influence. |
Spatial_preset | Random, left/right travel, expanding, contracting, alternating, Layer-based or Breathing Space. |
Preset dynamic ranges at 50% Compositional Intensity
| Preset | Dynamic Range | Dynamics Influence | Maximum possible conductor attenuation |
|---|---|---|---|
| Emergence | 8 dB | 0.50 | 4.0 dB |
| Accumulation | 8 dB | 0.50 | 4.0 dB |
| Breathing Form | 8 dB | 0.50 | 4.0 dB |
| Climax & Rupture | 8 dB | 0.55 | 4.4 dB |
| Spatial Counterpoint | 6 dB | 0.45 | 2.7 dB |
| Fragmentation | 8 dB | 0.45 | 3.6 dB |
| Controlled Chaos | 10 dB | 0.55 | 5.5 dB |
| Gesture Crescendo | 8 dB | 0.50 | 4.0 dB |
| Gesture Decrescendo | 8 dB | 0.50 | 4.0 dB |
These values are computed from the script's preset settings at intensity 50%. Actual gain reduction depends on the conductor curve and whether it reaches its minimum value.
Edit spatial layout: where each source sounds
The dialog opens after the folders have been scanned and the layers or streams have been established. It has one Position / Spread pair for each relevant region.
| Control | Default / range | Function |
|---|---|---|
Spatial_movement | 1; range 0–1 | 0 keeps the behavior inside source regions; 1 uses the whole field except for Layer-based. |
Position_1, Position_2, ... | −1 to +1 | Spatial center of each folder or stream, from left to right. |
Spread_1, Spread_2, ... | 0 to 2 | Permitted spread of event-center positions around the region. |
Stream_sources | Shared pool | Appears when one folder becomes several streams; choose shared files or split files between streams. |
Edit details: finer controls
| Control | Default | Function |
|---|---|---|
Duration_order_strictness / Duration_preference | 0.7 / 0.5 | Source-duration choice; preset values may override the first number. |
Gesture_awareness | Preset-dependent; Custom 0 | 0–1, soft musical-energy preference during source selection; 0 disables it. |
Randomness / Max_repetitions | 0.4 / 3 | Variation and source reuse limits; randomness may differ by preset. |
Min_silence / Max_silence | 1.5 / 6 s | Bounds for planned silences; some presets adjust them. |
Silence_placement / Ending_silence | Mode default / 0 s | Where pauses occur and whether the composition ends in silence. |
Dynamics_influence / Density_influence / Spatial_influence | Preset-dependent | How strongly the conductor controls each dimension. |
Max_correction | 12 dB | Full-mode library balance limit. |
Per_file_leveling / File_level_limit | Off / 4 dB | Optional bounded reduction of differences between files in a library. |
Output_sample_rate | 0 = auto | Most frequent source rate; highest breaks ties. |
Event_fade_in / Event_fade_out | 10 / 40 ms | Short anti-click fades at event boundaries. |
Final_fade | 0 s (Off) | Optional global ending fade; off by default to preserve natural source tails. |
Normalize / Peak_dBFS | On / −1 dBFS | Final peak scaling, separate from conductor gain. |
Memory_budget | 1024 MB | Threshold for disk-backed chunked rendering. |
Save_plan / Save_catalog | Off | Write optional metadata files rather than audio copies. |
Edit layers: the per-library dialog sets its duration preset, density factor, Gain_dB, rest probability and seed offset. To set Position and Spread, use Edit spatial layout instead.
What changed most recently: Allow_overlap_within_layers is in Edit composition controls, not in the Channels selector. Its default is Yes, regardless of preset, to preserve the earlier algorithm. The new Multitrack rendering option and Send Arrangement to Ableton Live are independent output capabilities.
16. Results and saved plans
| Output mode | What remains | When to choose it |
|---|---|---|
| Create in Praat (default) | Rendered Sound or, for memory-heavy output, disk-backed LongSound. | Hear and continue working on the finished composition in Praat. |
| Plan only | Plan and Settings Tables; no audio. | Inspect, save and reproduce the event score. |
| Export composition | Rendered mix and optional synchronized stems saved to files. | Move finished audio to another application. |
| Render a saved plan | Audio reconstructed from selected Tables or saved TSV plan/settings. | Return to a fixed arrangement, including its saved routing. |
| Profile source energy | Energy Table, figure, report and audition Sounds; no composition. | Listen to and check within-folder musical-energy rankings. |
| Export Arrangement JSON | A versioned, plan-only JSON describing complete events and source paths. | Transfer an editable score without rendering audio. |
| Send Arrangement to Ableton Live | JSON queued automatically in the local Live Bridge inbox; no audio rendered. | Populate Ableton Arrangement via the Max for Live importer. |
A composition plan records source paths, event entrances and durations, gain decisions, spatial positions, fades and silence information. It contains no copies of the audio. In Plan only, select the resulting Plan and Settings Tables and use Render a saved plan; alternatively, use a previously saved *_plan.tsv file with its accompanying *_settings.tsv.
Stereo, 4-channel and 8-channel renders retain their spatial channel distribution. Multitrack (one mono channel per layer) instead keeps compositional layers as separate mono channels of one Praat Sound, with their internal overlaps summed. For ordinary listening the script may create a temporary stereo audition; the multichannel original is retained. Additional Layer Stems are not generated in Multitrack because the channel separation is already present.
Memory_budget, the script renders in chunks to a WAV and opens a disk-backed LongSound. That WAV must remain available as long as the LongSound is in use. Successful in-memory renders do not need an output folder or an intermediate audio file.17. Editable arrangements in Ableton Live
TimeLine Composer can transmit its event score directly to a working Max for Live device, or save the same data as Arrangement JSON. This is not a WAV or stem transfer. The JSON refers to the original source files and records source identities, event onsets, complete durations, layer identities, planned gains, fades, spatial metadata and silences. Ableton uses the original recordings as editable Arrangement clips.
Send directly from Praat
Select Output → Send Arrangement to Ableton Live (background, no audio). On Windows, Praat writes a verified JSON file into LOCALAPPDATA/PraatAudioTools/LiveBridge/inbox. A Node for Max watcher in the Live device detects the file and asks the importer to place its events. There is no user file-selection step and no Python dependency. This is local file-based handoff, not a network message or remote server. Praat reports that the file was queued; the Max for Live device confirms the actual import.
One editable Audio Track per folder
- Select the source folders. For example, Piano, Violin and Percussion paths separated by semicolons.
- Open Edit composition controls. Set
Organization = Independent layers (one layer per folder). - Disable internal overlap. Untick
Allow_overlap_within_layers. Different folders can still play at the same time. - Choose Send Arrangement to Ableton Live. Keep
Channels = Stereo; it affects Praat rendering, not the number of Live tracks. - Inspect Live and the Praat report. With three actual layers, zero same-layer overlaps and a compatible importer, expect three ordinary Audio Tracks, one per source library, with independently editable clips.
When Allow_overlap_within_layers = Yes, the importer allocates extra voice tracks as needed to preserve overlapping source clips. Do not force all overlapping original clips into a single Live Audio Track: clips there cannot play independently at the same time. If you need exactly three tracks while retaining internal overlap, use a rendered layer-separated audio output instead of individual editable source clips.
Continuing the composition
The Ableton arrangement is an editable compositional scaffold, not a claim of identical sonic playback. The initial importer recreates the chosen recordings and their timeline positions; the composer is free to develop gain, panning, fades and processing in Live. The JSON retains richer planned metadata for documentation and possible future use.
18. How it works
1. Scan and profile
Read source durations, sample rates and channel counts; measure loudness and optional musical-energy descriptors. The source files remain on disk and are not altered.
2. Plan events and silences
Choose whole recordings using temporal preferences, folder-relative energy, repetition limits, source availability and the conductor. Plan global absolute silences and honor the optional non-overlap policy in each layer.
3. Shape the relations
Calculate entrances, density, levels and spatial positions; repair incidental gaps when possible without crossing internal silence or non-overlap restrictions. Let the final source tails finish naturally.
4. Render or transfer
Render stereo, spatial multichannel or layer-separated audio in Praat, or export / send the event plan as JSON for editable placement in Ableton Live. JSON transfer does not render audio.
The report includes actual event count, source usage, mean and maximum polyphony, measured silence, last audible end, library balance adjustments and several indicators of where the musical climax may occur. These are diagnostics, not substitutes for listening.
The script uses native Praat for composition and audio rendering. It does not time-stretch or arbitrarily cut source recordings. Natural Ending permits complete final tails after the nominal planning horizon; internal absolute silences remain protected. Arrangement export and the local Live Bridge transfer event metadata and source references, not copies of the library. No Python is required for the Praat workflow.
19. Notes and troubleshooting
- Can I compose from only one folder?
- Yes. Leave Folder paths empty and set Folders to browse to 1. You can also type one path directly. Accumulation and Spatial Counterpoint can derive three streams from the same folder.
- Why are there both a folder count and semicolon-separated folder paths?
- Folders to browse controls how many folder-picking dialogs appear only if Folder paths is empty. If you supply one or more semicolon-separated paths, those paths define the actual folders; the browse count is ignored.
- Why do my custom spatial positions have little or no effect?
- Open Edit spatial layout and inspect Spatial movement. Its default of 1 means full-field behavior takes precedence with all spatial presets except Layer-based. Choose 0 to keep events inside each group's region, or an intermediate value to mix fixed placement with global movement.
- Can the script identify which instrument is playing within a mixed folder?
- No. It organizes source files and source-folder identities, not instrument classes detected from audio. For reliable instrument-group placement, use separate folders or manually distribute files among streams.
- Where do I change the difference between soft events and the climax?
- Enable Edit composition controls in the main dialog and change Dynamic range (dB) beside the Conductor preset and Climax position. Edit details separately contains Dynamics influence. Their product sets the potential maximum conductor attenuation, not the measured amplitude range of the final mix.
- Why does the loudest passage not fall exactly at Climax Position?
- The conductor guides gain, density and spatial width, but source files differ in level, attacks and durations. Simultaneous events may also alter the actual peak. The conductor's climax is a compositional plan, not a guarantee of measured maximum loudness.
- Why is the composition quieter or less dense than expected?
- Compare the selected preset and Compositional Intensity with Event Density, Overlap and Dynamics Influence in the edit dialogs. A small library or low Max Repetitions can exhaust eligible source material before the desired climax.
- Why do I hear silence that I did not explicitly request?
- The engine tries to repair accidental gaps, but suitable full-length source recordings may not fit an available phrase. It reports remaining unplanned silence.
- Can I use different sample rates and stereo source files?
- Yes. Sources may use different sample rates. Praat renders to Stereo, 4 or 8 spatial channels, or Multitrack with one mono channel per compositional layer. This is distinct from Ableton track allocation.
- Does the script duplicate my audio library or create temporary WAVs?
- It leaves source files in place and does not copy them. In-memory rendering does not require temporary WAV files. Long, memory-constrained compositions may need a disk-backed WAV and a retained LongSound. Scratch chunks are cleaned after successful assembly; an unexpected crash or force-stop can bypass script cleanup.
- Can the same composition be reconstructed later?
- Yes. Save Plan and Settings metadata together with access to the original recordings. Saved plans include the routing and overlap policy where available. Older plans default to historical polyphonic / spatial behavior. JSON is a separate interchange format for editable arrangements.
- Where is the control that prevents overlapping clips on the same Ableton Track?
- Enable Edit composition controls, select Independent layers and untick Allow_overlap_within_layers. Its default is Yes. The setting changes event planning, not just audio routing. After exporting, check
Intra-layer overlaps: 0and the expected Live Audio Track count. - Why did I receive many Ableton tracks from only three folders?
- Several source recordings from the same layer overlapped. The importer used additional voice tracks so all clips could play simultaneously. To get one editable Track per folder, disable same-layer overlaps and use Independent layers. Multitrack in the Channels menu is unrelated.
- What is the difference between Multitrack and Send Arrangement to Ableton?
- Multitrack creates a single rendered Praat Sound with one mono channel per compositional layer (overlaps inside that layer are summed). Send Arrangement exports an editable clip timeline; track allocation is handled by Max for Live.
- Why does a 180-second composition last longer than 180 seconds?
- The target is a planning horizon. Recordings beginning before its end may continue naturally beyond it. The global Final fade is off by default. An explicit Ending silence is placed after the latest natural source tail.
- How do I hear whether the source-energy ranking is meaningful?
- Choose Output = Profile source energy, then examine the ranked Table and listen to EnergyAudition excerpts. During composition, Gesture_awareness in Edit details sets how strongly source-energy ranks bias selection.
- Why does Ableton sound different from the Praat render?
- The initial Arrangement importer preserves editable source identities and timing, not necessarily Praat's complete event gain, spatial panning, global normalization or fades. Continue shaping those interpretive details in Live.
For a direct workflow from three folders to three editable Ableton Tracks, see Non-overlapping layers and Editable arrangements in Ableton Live.