Selah 2.1 — Signal designer and ASIO shortcut

Selah 2.1 adds a visual serial signal designer inspired by hardware editor workflows, with colored modules, a selected-effect inspector, VST3 parameter access, drag reordering, local A–D tone snapshots, and tone undo/redo. A prominent ASIO settings button opens Audio & VST3.

Includes the existing self-contained CPU audio engine, ASIO I/O, VST3 effects hosting, cabinet IR support, tabs, sequencer and ChatGPT-account companion. No Python installation needed for the portable app.

Validation: production build, 9 Node tests and 7 native tests passed locally. Browser checks covered the new interactions and 390-pixel layout. Parallel routing and full Guitar Rig feature parity are not implemented. The native engine is unchanged from 2.0; the matching upstream source archive is attached.

Extract the full ZIP folder and open Selah.exe. AI features require the official Codex CLI and an eligible ChatGPT account; no API key. Unsigned Windows portable release. Third-party plugins are installed separately.

Selah 2.0 — Native ASIO, VST3 & Black Studio

Selah 2 adds native ASIO and VST3 processing and a configurable black studio workspace.

Windows x64: download Selah-windows-x64.zip, extract the entire folder, and launch Selah.exe. Python and the native audio runtime are bundled. The portable release is unsigned. Preserve your saved session/rack files when replacing an earlier version.

New in this release:

• Native CPU float32 DSP with ASIO device I/O, channel routing, sample-rate/buffer selection, driver-reported latency, DSP budget, clipping and driver-flag meters.
• 64-bit VST3 effect loading, order/bypass controls, exposed parameters, plugin-owned editors, isolated native audio process and restart recovery.
• True cabinet convolution using WAV/AIFF impulse responses, plus gain staging, a noise gate and final limiter.
• Native transport for the shared arrangement and local reference; processed guitar-sequence exports through the VST3 rack as 24-bit PCM WAV.
• Native rack presets with VST3 state and embedded cabinet IR audio.
• Black/graphite surfaces, custom accent color, interface scale, spacing, panel visibility and widths.
• Official developer links for free NAM, BYOD, Chow Matrix, Chow Tape Model and standard TDR Nova downloads. External plugins and neural model files are not bundled.

Start here: Audio & VST3 → select ASIO and your available interface/channels → start at 256 samples → Start audio → enable Guitar monitor after connecting headphones. Stop audio before loading plugins/IRs or exporting the native sequence.

Validation: 9 Node tests + 7 native tests pass; production build succeeds. ASIO4ALL opened at 44.1 kHz/256 samples and a generated guitar part was processed through TDR Nova 2.2.3 VST3. The short low-level test reported no driver flags or clipped blocks. Plugin parameters, native editor open/close, state restore, cabinet IR restore and native WAV export were exercised. The self-contained runtime enumerated ASIO without the development Python interpreter.

Driver-reported I/O latency is not measured analog round-trip latency. GPU DSP is not implemented. Physical guitar tone, every plugin and equivalence to a premium hardware pedal are not certified. Python coordinates streaming around compiled DSP; use a larger buffer if your chain misses deadlines. See the README and validation record.

The original UI/bridge remains MIT licensed; the separate native engine is GPL-3.0-or-later. Dependency notices and native source accompany the runtime. Selah-native-upstream-source.tar.gz supplies the exact Pedalboard release source with its pinned submodules and Selah native source/build script. ChatGPT features still use your eligible account through the official Codex CLI with no API key; plan access and limits apply.

Selah 1.0.0 — GuitarSim AI

Selah is a free, open-source local guitar workstation for worship practice.

Download Selah-windows-x64.zip, extract the entire folder, and open Selah.exe. The portable Windows x64 app includes its desktop runtime and is unsigned.

Included:

• Local song upload, waveform playback, tempo/key estimates, and editable single-note transcription drafts.
• Compression, drive, cabinet-style filtering/EQ, chorus, tempo-synced delay, convolution reverb, and worship presets.
• A shared guitar-tab editor and step sequencer, plus tab, MIDI, WAV, and session exports.
• Live guitar monitoring through an audio interface.
• An assistant using your existing ChatGPT account through the official Codex CLI, without an API key.
• AI live web search for free tab/chord resources, with separate song identity, version, access, and accuracy labels.
• AI-generated original practice parts that can be loaded into the sequencer and tabs.

For AI, install the current official Codex CLI and sign in using codex login. Your ChatGPT/Codex eligibility and usage limits apply. No ChatGPT subscription is bundled or sold by this project.

Exact pedalboard recovery and perfect full-mix guitar transcription are not promised. Source identity checks do not certify musical accuracy. Review the README and validation record.

Nine local automated tests, a production build, browser/audio-export checks, and live account-backed tone/source/arrangement requests completed. No private audio, sessions, or credentials are included.