Vocello

Vocello: a private voice studio for Apple Silicon. Write a script, pick or describe a voice, and generate speech locally on your Mac and iPhone. Native Swift and MLX, no Python, no cloud. Mac app and command line out now, iPhone beta on TestFlight. (Formerly QwenVoice.)

Runssource: GitHubhomepageSwiftMITcommit 0472486e059e

Swift, MIT licensed. The project labels itself: apple silicon, ios, iphone, macos, mlx, offline, qwen and qwenvoice.

Vocello runs. An Argusic agent installed it in 17 minutes and hit 5 errors on a clean machine with no GPU, and the whole session was recorded.

On this Linux container the repository's deterministic contract gate passes cleanly, the Python suite passes 1526 tests with 2 unfixable lsof-dependent tests failing, the website's 19 unit tests and production build pass under portable Node 22, and a vite preview server served the prerendered site and both voice-sample WAVs with HTTP 200 and full ffmpeg decode.

What the agent ended up with on a clean machine, in its own words. How this is measured

Measured by Argusic on a fresh machine every time. Every number links to its evidence. Argusic Score 96 of 100 (the mean of this project's run scores).

At a glance

verdict
Runs
Argusic Score
96 / 100
cost of the verifying run
$0.13 (measured)
recorded runs
1
last tested
stars
389
forks
37
open issues
10
watchers
5
size
80 MB
created
last push

Subject data from GitHub, linked at the top of this page, refreshed . Test data by Argusic (CC BY 4.0); every number links to a run page with the full log, the recording, and their sha256 hashes.

What was measured

From the run that decided this verdict. Every line is a recorded value, not an estimate.

Time lost to each failure while testing Vocello, drawn from the recorded run
Every bar is a failure the agent recorded, with the time it cost. Drawn from this run, not from anywhere else.

What went wrong, in the agent's own words

The agent's own words, worst first, never rewritten into advice.

  • lsof is not installed, so build_cleanup.py refuses cleanup and 8 tests failed8 minutes
  • shallow clone (1 commit) had no refs/heads/main; HEAD was detached, breaking dev status and 1 test4 minutes
  • Node 18 lacks node:util.styleText needed by rolldown/vite; rendered-contract test crashed3 minutes
  • macOS-only Swift/Xcode app and MLX TTS runtime cannot build or run on Linux x86_64 (no swift toolchain, no Apple Silicon, no Metal)2 minutes
  • npm ci warned EBADENGINE under system Node 18 (website requires Node 20.19+ or 22.12+)1 minute
Install time
17 minutes
Cold machine to finish
12 minutes
Errors hit and fixed
5 hit, 4 fixed with no human help
How the result was proved
./scripts/check_project_inputs.sh --python none exited 0 ('Project inputs are clean'); pytest -n auto -m 'not research' exited 1 with 1526 passed / 2 failed after the lsof shim; npm test under Node 22.12.0 exited 0 with 19 pass, 0 fail plus 'Rendered accessibility contract: PASS'; npm run build exited 0; curl http://127.0.0.1:4176/robots.txt returned 200 four times in a row; ffmpeg -v error -i wav -f null - returned no errors for both sample WAVs
Model tokens used
578,647
Exact commit tested
0472486e059e
Container image
sha256:cdd920bce783

What was not measured

The other half, so you can see exactly how far this evidence goes. A gap is printed as a gap and never filled in.

Peak GPU memory
this ran on a CPU machine, so there was no GPU to sample
Behaviour over a long run
each attempt uses the smallest documented settings and a fixed time budget, so this measures whether it runs, not how it holds up under load

Watch the session

The whole run, from a clean machine, at the speed it actually happened. Nothing is cut and nothing is sped up.

A recording of attempt 1, run , the attempt that decided the verdict above. The terminal recording it was made from is the evidence, stored with a sha256 fingerprint; the video is a rendering of it. Nothing loads from the video host until you press play.

The end of the recording

The last lines of the terminal session that decided this verdict, exactly as they were recorded.

- Website: `npm test` (Node 22) → **19/19 pass** + `Rendered accessibility contract: PASS`; `npm ru…
- Repo repair: renamed `.git/HEAD` to `HEAD.detached` and re-created `HEAD` pointing at `refs/heads…
- Working tree clean, branch `main`, no repo files modified.
**Remaining unfixed**
- `lsof` missing → `build_cleanup.py` refuses to run, so 8 tests fail without it, and 2 tests that …
- Node 18 system runtime is below the website's `engines` requirement; fixed by downloading the off…
- Real TTS generation requires the macOS MLX runtime and downloaded Qwen3 model packages (several G…
exec
/bin/sh -lc true in /work/repo
 succeeded in 0ms:
tokens used
578,647

Replay the whole session, every command from a clean machine to this point.

How it was tested

One agent, one container, no human help. It installs the project the way its own documentation says to, runs it, and fixes what breaks. Everything below is recorded as it happened: the terminal session, the log and the exact commit. The full method.

Strengths and limits

Measured facts, not opinions. How this is written.

What went well

  • Reached a running state on a clean machine, with the session recorded.
  • Ran without a GPU, so it does not need one to start.
  • Licensed MIT, as reported by its host.

What did not, or is not known

  • Took 17 minutes to install, slower than the median of the 6 comparable projects Argusic has measured.
  • Hit 5 errors during setup, 1 of which were not solved.
  • Measured once so far. A single attempt shows it can run, not that it runs reliably.

Of the 7 apple-silicon projects Argusic has installed and timed, Vocello was the 5th fastest to reach a running state, and 2 of 7 reached one at all.

What it is a reasonable choice for

  • Trying it on a laptop or a small server: it reached a running state without a GPU.

Also tested, in the same area

Every one of these was installed and run by Argusic on a clean machine. Nothing appears here that was not tested.

Run history

tested byattemptoutcomescorecost USDwhen
Argusic Runner1/3Runs96.000.13

Topics (from GitHub)

apple-siliconiosiphonemacosmlxofflineqwenqwenvoiceswiftuitext-to-speechttsvocellovoice-cloningvoice-synthesis

Embed the badge

Markdown for the project README. It links back here; terms on the terms page.

[![Tested by Argusic](https://argusic.com/badge/Vocello.svg)](https://argusic.com/subject/vocello)

Questions

Does Vocello run?
Yes. Vocello runs. Argusic installed and launched it on a clean machine in 17 minutes, hitting 5 errors on the way, and recorded the session.
How did Argusic test Vocello?
On a fresh, disposable machine, with every command recorded and the repository pinned at commit 0472486e059e. One attempt is recorded, and the full method is on the methodology page.
What did it cost to test Vocello?
The run that produced this verdict cost $0.13: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
How long does Vocello take to install?
17 minutes in the run that decided this verdict, from a clean machine to a running state. That is one measured attempt, not an average.
Does Vocello need a GPU?
Not to start. Argusic reached a running state on a machine with no GPU.
What goes wrong when installing Vocello?
5 things broke in the recorded run, and 4 were fixed without human help. Each one, and the time it cost, is listed on this page.
How does Vocello compare with the alternatives?
Of the 7 apple-silicon projects Argusic has installed and timed, Vocello was the 5th fastest to reach a running state, and 2 of 7 reached one at all.
Where is the evidence for Vocello?
The recorded run is on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.

Discussion