Ressoar

A small, private web app for practicing voice feminization. It guides you through drills, runs a live pitch graph while you speak, shows your camera so you can watch your mouth and jaw, and lets you keep recordings of your own takes. It is built for trans women and anyone working toward a more feminine voice, at their own pace, with no account and no data leaving the device.

The name comes from the Portuguese word for "to resonate." Resonance, not just pitch, is what makes a voice read as feminine, and this tool is meant to help you hear and see both.

What it does

• Guided drills. A session is a list of exercises, each broken into timed steps (for example, a 90-second glissando). Start a step when you are ready and a timer counts down.
• Live pitch tracking. While you speak, a graph plots your pitch in real time across the 80-400 Hz range, with the current value shown in Hz and as a musical note in both English (A3) and Portuguese solfege (Lá3).
• Camera feedback. Your webcam is mirrored next to the graph so you can watch your articulation, smile, and jaw position.
• Recordings you control. Each take is recorded (video and audio) and downloaded to your device when you stop. Nothing is uploaded anywhere.
• Checklist progress. Mark off the steps you have completed. Progress is saved in your browser per session.
• Your own sessions. Load a session from a file, or build one in the editor and download it to reuse and share.
• Portuguese and English. The interface defaults to Portuguese; use the flag in the top-right corner to switch to English.

A note on privacy

Everything runs in your browser. Your camera and microphone are used only to draw the live graph and to create recordings that download straight to your device. There is no server storing your audio, video, or progress. If you host the app yourself, the same is true: the deployed copy serves static files only.

Using the app

1. On the home screen, choose Load Session, Create Session, or open the built-in Glissando example.
2. In a session, each step has a checkbox and a Start drill button.
3. When you open a drill, allow camera and microphone access. Press Ready? Start to begin. The timer runs, the pitch graph draws live, and recording begins automatically.
4. When the timer ends (or you press Stop), the recording downloads to your device. Use Download recording to save it again if needed.
5. Check off steps as you finish them. Your progress is remembered the next time you open the same session in the same browser.

A few practical notes:

• This tool shows pitch. Pitch matters, but resonance, intonation, and weight matter at least as much. Use the camera and your own ear alongside the graph, not the number alone.
• Warm up gently and stop if anything hurts. Strain is never the goal.

Session file format

Sessions are plain JSON files with a .ressoar.json extension, so they are easy to read, edit, and share. A duration of null means an untimed ("free") step.

{
  "title": "My session",
  "createdAt": "2026-05-23",
  "exercises": [
    {
      "title": "Sustained vowel",
      "description": "Optional notes shown above the steps.",
      "steps": [
        { "label": "Hold an /i/ at a comfortable high pitch.", "duration": 90 },
        { "label": "Read a paragraph aloud, twice.", "duration": null }
      ]
    }
  ]
}

The Create Session editor produces exactly this format and downloads it as <title>-<dd-mm-yyyy>.ressoar.json. You can keep personal sessions outside the project (this repo ignores a saved-sessions/ folder for that purpose) and load them with Load Session.

Running locally

This project uses mise to pin the toolchain (Bun) and serves the public/ folder as a static site.

mise install        # installs the pinned Bun version
bun run dev         # starts the server with hot reload

Then open the printed URL (default http://localhost:3000). The camera and microphone work on localhost without HTTPS.

Deploying

The app is fully static, so any static host works. A Vercel configuration is included (vercel.json); it serves the public/ directory.

1. Push the repository to a Git provider Vercel can read.
2. Import the project in Vercel. No build command is needed; the included config sets the output directory to public.
3. Deploy.

Important: the camera and microphone require a secure context. They work on localhost, and on any deployed site they require HTTPS (Vercel and most static hosts provide this automatically). They will not work over plain HTTP.

Credits

Typeface: Monaspace Radon by GitHub Next, used under the SIL Open Font License.

Disclaimer

Ressoar is a practice aid, not medical advice. Voice training carries some risk of strain if done aggressively. If you can, work with a speech-language pathologist or a qualified gender-affirming voice trainer, especially when starting out. Be patient and kind to yourself; voice change takes time.