# Speak Savvy

Practice speaking on iPhone and get specific feedback on structure, pace, clarity, and delivery.

Status: active

## Speak Savvy gives each practice session a next step

Speak Savvy helps people rehearse short answers and see where they can improve. Eight drills cover quick practice, conversation, PREP answers, roleplay, explanation, stories, metaphors, and paraphrasing. After a recording, the app shows the transcript beside feedback on content and delivery. A speaker can spot a missing part of an answer, a pace spike, or repeated filler words before trying again.

## The iPhone measures delivery from the recording

Apple Speech transcribes the answer and supplies word timing and confidence. On-device signal processing measures pitch, volume, and voice quality; the app also calculates pace and filler frequency. Color-coded transcript segments show where speed changed. These measurements give the speaker specific evidence to replay rather than a single impression of how the answer sounded.

![Speak Savvy practice feedback ::border](/images/projects/speak-savvy/prep-feedback.jpeg)

## Coaching interprets the answer without replacing the measurements

Speak Savvy scores six dimensions: clarity, relevance, fluency, pace, delivery, and conciseness. The home view groups feedback into Structure, Clarity, Flow, and Inflection. Engagement remains qualitative instead of receiving a numeric score. Apple Foundation Models can add coaching on supported devices; when that model is unavailable, the app still gives guidance from measured signals.

The scoring paths have different jobs. Audio analysis computes repeatable delivery measurements. Coaching explains what those measurements and the transcript mean for the next attempt. Practice recordings, transcripts, scores, and history stay in SwiftData on the device.

## Release builds keep practice data on the device

Speak Savvy requires iOS 26. Release builds do not send practice data to a model provider or require an account. Cloud model experiments and live voice are limited to a separate debug owner panel; they are not part of the release experience.

## Links

- [HTML project page](https://rosslabs.ai/projects/speak-savvy/)
- [Demo](https://rosslabs.ai/images/projects/speak-savvy/screen-tour.mp4)