Overview
Module: Module 4: Local Voice Assistant
Activity type: Quiz / retrieval practice
Target: ≥80% before topic mastery unlock
Objective: Given CPU/RAM constraints, the learner can deploy local Whisper (STT) and Piper (TTS) via Wyoming, benchmark them separately, and record latency/quality tradeoffs before integration.
Instruction
Instructions
In plain words: This is the teaching core. One major concept (or a tight group). Read carefully.
More detail: Use What this means under dense blocks.
Close the reading. Write answers from memory. Then self-score.
In plain words: Big idea: Close the reading.
Learn
Questions
In plain words: Read this block, then try the next action. If you feel stuck, tap What this means.
Active recall: write answers without rereading first. Target ≥80% before the gate.
In plain words: Big idea: Active recall: write answers without rereading first.
- What does Wyoming provide versus Whisper?
- Why record cold and warm latency?
- What does a reachable TCP port fail to prove?
- Why keep Piper model and metadata together?
- How can an LLM interfere with STT even when both work independently?
- Which variable should change in a controlled comparison?
In plain words: A port is the door number you use in the browser to open the app.
Learn
After scoring
In plain words: Read this block, then try the next action. If you feel stuck, tap What this means.
- Missed items → return to Reading weak sections → redo Feynman Retry → reattempt missed questions.
- Passing score unlocks marking this topic complete on the module hub (still need lab gate + homework).
In plain words: Big idea: - Missed items → return to **Reading** weak sections → redo **Feynman Retry** → reattempt missed questions.