ToneTrace press kit

Product description

One line

ToneTrace is a desktop app that tags a producer's own sample library by instrument, mood and genre, lets them search it by feel, and lifts individual sounds out of longer files.

Short

ToneTrace listens to a producer's own sample library and tags every file by instrument, mood and genre, so sounds can be found by how they sound rather than by their file names. A mood and energy pad narrows a library by feel, and because ToneTrace records where each sound occurs inside a file, a single moment can be dragged out as its own WAV. Two open audio models run on the user's machine by default; nothing is uploaded in the current build. It runs on macOS, Windows and Linux, and beta applications are open.

Long

ToneTrace is a desktop application for music producers, sound designers and educators who work from large sample libraries. It listens to each audio file with two open models, YAMNet from Google for instruments and sound events and CLAP from LAION for mood, genre and production character, and tags the file with the models' own confidence figures. Every file is fingerprinted so nothing is analysed twice, even after being moved or renamed. Users search names and tags together, ask for sounds similar to one they like, audition with a built-in player, drag files straight into any DAW as native file drags, and can write the tags into the files themselves as standard metadata. Processing is local by default: both models run on the user's own computer through ONNX Runtime, and the current build uploads nothing. An optional faster processing path is being evaluated for launch. ToneTrace runs on macOS (Apple Silicon and Intel), Windows and Linux, and beta applications are open ahead of a public launch.

Facts

Product

ToneTrace, desktop application

Website

https://tonetrace.app

Status

Pre-release; beta applications open

Platforms

macOS (Apple Silicon and Intel), Windows, Linux

Built with

Electron, React, ONNX Runtime, SQLite, FFmpeg

Models

YAMNet (Google, Apache 2.0), CLAP (LAION, Apache 2.0)

Notable

Mood and energy pad; per-sound regions with micro-sample export

Processing

Local by default; nothing uploaded in the current build

Pricing

The beta is free. Pricing will be announced at launch.

Logo files

SVG files are the masters. PNGs are provided at common sizes. The mark is a pale indigo tile with an indigo glyph of one continuous stroke tracing a waveform; the wordmark is set in Lexend.

• Keep clear space around the mark equal to the stroke width of the glyph.
• Do not recolour the tile or the glyph, rotate the mark or set the wordmark in another typeface.
• On light backgrounds use the dark wordmark, on dark backgrounds the light one; the mark itself is the same on both.