A quiet place to practice speaking

Rehearse the conversation before it happens.

Scenarios you write, voices that respond, transcripts you can revisit. Talk Buddy is a studio for the conversations that matter — the difficult ones, the first ones, the ones you want to get right.

Local-first. Your recordings and transcripts stay on this device.

How practice works

01

Pick a Scenario

A job interview, a difficult meeting, a client call. Choose from the built-in library, import a Practice Pack from a teacher, or write your own.

02

Have the Conversation

Speak naturally. Talk Buddy listens, and an AI partner replies aloud — turn by turn, like the real thing. Interrupt it whenever you want.

03

Revisit the Session

Every attempt is saved as a transcript with optional analysis, so you can see how you did, spot habits, and practice again.

Three parts, all yours to swap

Listening

Speech-to-text that turns your voice into words. Runs fully offline with the bundled Whisper server — no audio ever leaves your machine.

Voice

Natural text-to-speech that speaks the replies. The bundled Piper voices work offline; plug in a Speaches server for premium Kokoro voices.

AI Brain

Bring your own AI: Anthropic, OpenAI, Gemini, Groq — or keep everything local with Ollama. Your key, your choice, one settings tab each.

Privacy-first, by default

Talk Buddy is built for classrooms, counselling practice, and nervous rehearsals — the conversations you don't want on someone else's server.

For learners

Practice interviews, presentations, and awkward conversations at your own pace, in private, as many times as you need. Track your progress across sessions.

For teachers

Write scenarios that feel real, group them into Practice Packs, and share them with students as a single file — or a link. No accounts to manage.

Full guides for students, teachers, and scenario authors live in the documentation.