Inspiration
I’ve tried learning accents before and it always felt like guessing. You can listen to native speakers all day, but it’s hard to tell what you should change. I wanted something that makes the target obvious, fast.
What it does
MirrorAccent is an accent practice app. You record yourself, pick an accent, and it plays back your voice transformed into that accent. Then it gives you some nerdy but useful feedback (pitch, vowel/formant stuff, volume, speed, timing) so you know what to tweak.
How we built it
It’s an Electron desktop app. The “make me sound American” part uses ElevenLabs voice changer (Speech-to-Speech). The prosody analysis is powered by custom logic: pitch tracking (YIN), vowel/resonance comparison (LPC/formants), plus intensity and speaking-rate calculations.
Challenges we ran into
- Setup stuff (mic selection + ElevenLabs key +
speech_to_speechpermissions) was always more annoying than it should have been.
Accomplishments that we're proud of
- Hearing the “goal” in your own voice is honestly the best part; it clicks immediately.
- The prosody scores are actually helpful instead of just hand-waving it.
- It’s simple: record → transform → learn → repeat.
What we learned
If you can’t hear what “right” sounds like for you, you’ll waste a ton of time.
What's next for MirrorAccent
- Progress tracking so you can see improvement over time.
- Integration with open-source voice cloning tools.
- More accents + better reliability in noisy real-world setups.
Built With
- css3
- electron
- elevenlabs-s2s-api
- html5
- javascript


Log in or sign up for Devpost to join the conversation.