pebblab

Real speech · your voice · a little melody

echos.

Listen. Repeat.
Match the melody.

You can know every word and still miss the music. Hear a real speaker, repeat the phrase, and see how the rise and fall of your voice compares.

Free, open source, and right in your browser.

Echo practises the melody of speech with a glowing Cadence Sphere in a forest
Not their voice.Your melody.

01 / From your ears to your voice

Listen Repeat Compare

  1. 01

    Listen.

    A real person. A real phrase. Listen before the words appear, so your ears get the first turn.

  2. 02

    Repeat.

    Record what you heard. Your voice makes its own line: the melody you actually spoke.

  3. 03

    Compare.

    Look at the two contours. Hear the reference and yourself in turn. Notice a difference? Record again.

Each line is normalized to its speaker’s own pitch range and to the phrase’s duration. You’re comparing the shape, not whose voice is higher. No score. No “correct” voice.

ListenLet your ears go first.
Working Echos app in Listen state: a green reference pitch contour, hidden transcript, and Play reference control
RecordNow find it in your own voice.
Working Echos app in light theme with the US speech set selected, actively recording: Stop recording button and Recording status, with no learner comparison displayed yet

Captured from the working app, without interface edits. The Record view uses a test microphone; no learner results are simulated.

Make the next line yours — try Echos
DaCo Mozzie offers the Cadence Sphere to Echo at the mouth of a cave

02 / The idea began as a fable

the found melody

Echo could repeat the words. But she had lost their music. Then a collector of cadences offered her something stranger than a new voice: a way to find her own.

“You must learn the shape of the pitch, and then pour your own life into it.”

A story of Echo, DaCo Mozzie, and the Cadence Sphere.

Read The Found Melody

03 / Real voices make this possible

Shared speech.
New possibilities.

The demo uses 34 selected recordings from Common Voice Spontaneous Speech 4.0 – English, available through Mozilla Data Collective: 17 UK and 17 US clips.

People contributed their voices. Echos turns those real, spontaneous phrases into an opportunity to listen, notice, and practise.

Explore Mozilla Data Collective

Echos is an independent project. Mozilla Data Collective is the data source, not an endorsement.

Mozilla Data Collective homepage: Create. Curate. Control. A platform for sharing and discovering datasets.
Mozilla Data Collective website, shown for source context. Screenshot supplied for this project; website imagery and branding belong to their respective owners.

04 / Your browser. Your practice.

Your voice stays
on your side.

The app estimates pitch and draws both contours in your browser. It does not upload your recording or send it to a speech-to-text service. Your take is held temporarily, not saved as an account or a score.

The live demo needs microphone permission to record, and HTTPS or localhost. Local processing is a design choice, not a blanket security guarantee; use a browser and device you trust.

No score to chase. A pattern to notice.

Find the melody.
Keep your voice.

Try Echos

Opens the working browser demo on pebblab.com.