backchannel
Windows x64
For Windows 10 or later on 64-bit Intel or AMD processors.
- Download
Backchannel-windows-x64-v0.6.2.zip. - Unzip it and run
Backchannel.exe. - If SmartScreen appears, choose More info, then Run anyway.
Release v0.6.2
Starting a session no longer hangs.
Creating a new session could stick on "Creating..." and never finish. Nothing was wrong with the session: the app was quietly downloading a model in the background, that download jammed, and everything that stores text queued behind it. Saving text no longer waits on a download at all. Downloads are visible now too, with progress, a reason when one fails, and a Retry button. Local transcription, which a packaging mistake had broken outright in the desktop app, works again.
Download Backchannel v0.6.2
These unsigned open-source builds are delivered through the Backchannel download portal. Downloads are open to everyone; no account or GitHub sign-in is required.
The Windows and Linux bundles include ffmpeg. On macOS, voice calibration and compressed audio imports still need ffmpeg on PATH.
macOS Apple Silicon
For Apple Silicon Macs. Intel macOS builds are not included in this release.
- Download
Backchannel-macos-arm64-v0.6.2.zip. - Unzip it and move
Backchannel.appwherever you keep apps. - Right-click
Backchannel.appand choose Open the first time.
Linux x64
A portable bundle for 64-bit Intel or AMD Linux systems.
- Download
Backchannel-linux-x64-v0.6.2.tar.gz. - Extract it with
tar -xzf Backchannel-linux-x64-v0.6.2.tar.gz. - Run
Backchannel/Backchannel.
What changed
v0.6.2 is a fix release. A background model download could deadlock and take session creation down with it, and the same packaging fault broke local transcription.
Saving text no longer waits on a download
Creating a session protects the name you gave it, which reached the PII Shield's on-device model. That loader held a single lock across the whole download, so when the download jammed every protected write in the app queued behind it and never returned. Nothing on that path downloads or waits for a download any more. If the model is missing, still arriving, or failed, the shield protects text with the layers it already has and the model joins in once it is ready.
The download that jammed cannot jam that way again
It was never the network. A desktop window has no console, so the progress bar was writing to a stream that does not exist, and it took a process-wide lock before doing so and released it only afterwards. The error escaped with the lock stuck shut, and the next download blocked on it forever. Progress is now reported by something that holds no locks and writes to no stream, and the desktop build is given real output streams before anything can write to them.
Downloads say what they are doing
Backchannel fetches model weights the first time something needs them, and it used to do that with no sign anywhere. A notice now follows you through the app showing what is downloading, how far along it is, and what it is for, because downloads start from places you are not looking at: a warm-up at startup, the first local transcription of a call. A failure says what went wrong and offers Retry instead of leaving a feature looking broken for no stated reason.
Local transcription works in the desktop app
The bundle left out a small metadata file that one of the speech libraries reads about itself at startup, so every local Whisper and Parakeet transcription failed the moment it was asked to run. Both work in this build.
Download now returns instead of looking stuck
The Privacy tab's button for the name-recognition model waited for the entire transfer before responding. It now starts the download and comes straight back, with the progress shown underneath it.
First-run notes
| Need | What to do |
|---|---|
| Release access | Nothing - downloads are open to everyone; no account or sign-in is required. |
| Provider key | Follow the guided setup on first launch, or add a key later in Admin -> Connections. Calls will not start until the transcription model has a usable credential; local ONNX transcription needs none. |
| Voice calibration | Works out of the box on Windows and Linux. On macOS, install ffmpeg and make sure it is on PATH before enrolling your voice. |
| Upgrading | An install on v0.5.3 or later updates itself in place, with no account anywhere on the path: it verifies the new build's signature first and waits for recording and post-processing to finish before restarting. If a v0.6.1 install is currently stuck on "Creating...", quit it from the tray and start it again, then update. An install on v0.5.2 or earlier will not see this release from inside the app: download the bundle above, quit the old version from the tray, and run the new one. Session data is stored per-user and shared by every version, and the PII Shield stays off until you turn it on in Admin -> Privacy. The Admin About tab shows a what's-new notice after upgrading. |
| Desktop launch | Backchannel opens a dedicated local app window in your default browser when it supports app windows, otherwise it uses a standard browser tab. It remains available from the system tray or menu bar. |
| Docker | Check out the v0.6.2 source tag and rebuild Compose to run the same release through containers. If you reach the stack by a DNS hostname rather than an IP address or localhost, set BACKCHANNEL_ALLOWED_HOSTS to that name in .env first. |
Start with the desktop app or self-host the stack
Both paths run Backchannel on your machine. Choose the desktop bundle for the fastest start, or Compose for maximum control.