Threshold Lab Audio Laboratory

Product tour · Free in U.S. stores for Mac and Windows

Focused inner exploration.
Six connected workspaces.

Threshold Lab is built for self-directed exploration of focused altered states of consciousness through binaural audio and guided voice narration. Begin with a finished experience, study why it was designed that way, adapt an inspectable plan, or author the complete timeline by hand.

Tour the workspaces Open the learning center
20Included design recipes
40–1200 HzCarrier range
0.5–40 HzBeat-difference range
Up to 12Local profiles

Composition workflow · macOS

One composition to shape, inspect, and refine.

Follow a session from your first intention through its layers, proposed changes, and the version you actually hear. The new workflow keeps those decisions connected.

Available for Mac on the Mac App Store. The composition capabilities in this section are for the Mac edition. The Windows edition remains available separately; no Windows composition update is announced here. Get Threshold Lab for Mac.

01 · Shared editing

Keep every view on the same session.

AI Lab, editable Rows, Automation, playback, and WAV export use the same composition. Accepted changes carry through to the sound you render.

02 · Component inspection

See what is sounding at a chosen moment.

Name components and inspect their actual left/right frequencies, gains, fades, and placement. Temporarily mute or solo a component, or loop up to 30 seconds, without changing the saved version or ordinary export.

03 · Reviewed revisions

Ask for a focused change. Decide what stays.

Select a passage or component, protect passages you want to keep, and review the proposed before-and-after values. Accept, reject, or undo a change. A delayed AI response cannot replace a different session you have opened.

04 · Recoverable work

Explore a change without losing your starting point.

Undo and Redo survive reopening. Each composition retains its baseline and latest 49 revisions. At the 200-session capacity, new creation pauses instead of removing older saved work.

05 · Exact versions

Keep the design connected to the audio.

A saved version retains its settings, sound-generation seeds, renderer choice, prompt history, and references to exact local audio. Optional Experience Records can identify the version actually heard. Imported recordings remain attached through ordinary timing edits.

06 · Local output

Audition and export from the same audio engine.

Playback, short auditions, and WAV export share the audio rendering path. Desktop exports offer progress, cancellation, and recovery. Earlier compositions keep their original renderer unless you explicitly choose a reversible upgrade.

Exact version files reference audio stored on the same computer; they do not embed a self-contained audio pack. Keep the associated local assets when saving your work. AI planning supports sessions from 5 to 120 minutes and uses your chosen provider when requested. Manual editing and local revisions remain available without an AI request.

Continue to the workspace tour

macOS 1.0.5 screenshots

The feature list is extensive. The application makes it concrete.

Threshold Lab keeps advanced controls visible without turning every task into the same screen. These macOS 1.0.5 captures show the manual workbench, the Library, and profile separation at full application scale. They predate the composition controls described above.

Session Creator
Threshold Lab Session Creator showing The Witness at the Threshold as 19 editable rows with transport, voice, mix, and timeline controls.
Session Creator Edit every segment directly while the complete session structure remains visible.
Session Library
Threshold Lab Session Library showing The Witness at the Threshold ready to listen at 45 minutes, with the complete first read-only timeline rows visible and no right-hand rail.
Session Library Play an included session and inspect its complete read-only Rows timeline without clipping.
Local Profiles
Threshold Lab Local Profiles showing profile statistics, profile switching, and editable identity fields stored locally on the Mac.
Local Profiles Separate local workspaces keep current sessions, history, appearance, and settings distinct.

The application map

A clear place for every stage of the work.

Workspaces stay distinct so listening, planning, authoring, education, and reflection do not collapse into one complicated screen.

01Protocol

A repeatable way to learn from listening.

Protocol explains a practical loop: orient, calibrate, choose a baseline, listen, observe only if useful, change one meaningful variable, and return safely.

  • Comfort and headphone calibration
  • Comparison-first experimentation
  • Reliable return as a design requirement
02Education

Science and a complete user guide inside the app.

Seven Learn chapters explain audio perception, frequency vocabulary, sound beds, voice guidance, session architecture, evidence, and safety. Fourteen guide chapters cover every workspace and advanced field.

  • Worked binaural-frequency examples
  • Noise color and nature-bed reference
  • Advanced tone-layer field guide
03AI Lab

Natural-language planning without an opaque result.

Describe the intended session in ordinary language. A supported provider proposes a bounded plan, then Threshold Lab validates and displays its audio, narration, and protocol timelines before local rendering begins.

  • Built-in example based on The Witness at the Threshold
  • Generate, clear, import, and export plan controls
  • Local voice rendering, audition, and WAV export

Provider credentials belong to the user. The provider proposes a structured plan, not companion chat; the app performs narration and procedural audio rendering locally.

04Session Library

A listening library that also teaches construction.

Twenty original design recipes span eight families. A fresh installation writes all twenty authored prompts, twenty exact JSON plans, and twenty descriptions into the local Library. It also includes the finished 90-minute Witness WAV for immediate playback and persistently identifies the other nineteen designs that still need local rendering.

  • Twenty original recipes with sixty readable supporting files
  • One complete Witness WAV ready to play immediately
  • Persistent progress plus Open Next in AI Lab for the remaining nineteen
  • Independent transport for ready audio: play, pause, stop, previous, and next
  • Prompt, plan, and description inspection before rendering
  • Direct opening in AI Lab for adaptation or rendering
05Experience Records

Optional observations tied to the exact session version.

Record body rest, awareness continuity, return quality, free-form observations, and a possible next change. Every field is optional.

  • Readable local JSON records
  • No streaks, grades, required entries, or locked content
  • Direct link back to the related session version
06Session Creator

The advanced manual workbench.

Start from a guided preset, an explorer preset, imported text, a session recipe, an AI plan, or external audio. Then edit the complete timeline and signal architecture.

  • Editable, reorderable timed segments
  • Stacked tone layers with per-layer routing, offsets, gain, and fades
  • Voice, noise, nature, binaural, and master mix controls

Session Creator deep dive

Every important audio choice remains explicit.

A segment may carry as many as eight independent tone layers. Each layer has a carrier pair, beat-difference journey, gain, channel routing, active window, and edge fades.

Tone, noise, nature, voice, and master remain separate buses, making it possible to preserve narration clarity and headroom without rewriting the timeline.

ControlRange or choicesWhat it changes
Carrier center40–1200 HzAudible center pitch
Beat difference0.5–40 HzLeft/right frequency separation
CurveHold or linearFixed value or gradual journey
RoutingBoth, left, rightStereo pair or advanced single-channel texture
Layer windowOffsets + fadesWhere a layer enters and leaves its parent segment
Voice speed.75, 1.0, 1.25Local narration delivery

Five independent buses

Layered by design. Balanced by ear.

The app reports digital gain, but it does not pretend to know the sound pressure at the ear. Headphones, system volume, fit, and hearing remain part of the complete playback chain.

01

Binaural / Tone

All generated tone pairs and stacked layers.

02

Noise

One of seven procedurally shaped noise colors.

03

Nature

Ocean, rain, waterfall, stream, or none.

04

Voice

Rendered or imported narration with stereo pan.

05

Volume / Master

The complete output with session-edge fades.

AI-assisted, user-directed

Planning is online only when you request it. Rendering stays local.

Connect ChatGPT through Threshold Lab, or supply an OpenAI or xAI API key in Settings. The Mac edition keeps provider credentials in macOS secure storage. No AI subscription or provider usage is bundled with the application.

Review the complete data boundary
  1. 1
    Your requestIntention, duration, style, and constraints
  2. 2
    Selected providerReturns a proposed structured plan
  3. 3
    Local validationChecks limits, timing, fields, and return structure
  4. 4
    Local renderingVoice, tones, beds, playback, and WAV export

Learn from complete examples

The Library reveals the source materials before you render the sound.

The included collection contains twenty original recipes across eight design families. Their sixty prompt, plan, and description files are available immediately. Witness also includes its fourth matching artifact—a finished lossless WAV. Each local render completes one of the remaining nineteen learning bundles.

Calibration Foundation Witness Silence Study Return Confidence Loving Awareness Threshold Variability Integration
  • TXTPromptWhy the session was requested
  • JSONPlanHow the session is timed and built
  • TXTDescriptionWhat the finished plan does
  • WAVLossless masterIncluded for Witness; rendered locally for the other nineteen

A shared app with personal workspaces

Profiles organize the work. Themes make the laboratory comfortable.

Up to twelve local profiles can keep separate current sessions, pinned items, history, voice choice, and saved AI Lab state on one Mac.

Profiles are an organization tool, not an operating-system security boundary. The installed Library, exported files, Experience Records folder, appearance, and AI connection remain shared within the same Mac login.

JW AR MK SL
Light Dark System Custom

What the first edition is designed to deliver

A serious creative instrument that still teaches the new user what each control means.

Original learning library Built-in science education Fourteen-chapter user guide AI-assisted plan creation Direct manual authorship Local narration rendering Lossless WAV export Readable companion files Optional Experience Records Up to twelve local profiles Light, dark, system, and custom themes Clear safety and evidence boundaries

Continue the tour

Understand the technology before deciding how to use it.

The learning center explains binaural perception, frequency vocabulary, noise colors, voice-guided design, evidence limits, and listening safety.

Enter the learning center