Drop an episode and get timestamps, chapter titles and show notes
Intermediate16h1–3
WhisperEmbeddingsLLM
16/20Customer Call Sentiment Timeline
See where in a call the customer got frustrated and why.
Intermediate16h1–3 people16/20 fit
Build Customer Call Sentiment Timeline for support and sales managers in ~16h with 1–3 people. Stack: Whisper + LLM + Charts + Next.js.
Why it wins: The timeline visual makes a complex analysis instantly readable.
Build flow
01 · Problem
Managers cannot listen to every call and miss the exact moments that lose customers.
02 · Solution
Transcribe calls with speaker labels, score sentiment per turn and plot a timeline with the turning-point quotes highlighted.
03 · Build
- 1Transcribe with timestamps
- 2Split turns by speaker
- 3Score sentiment per turn with the LLM
- 4Plot a clickable timeline
04 · Demo
Play a sample call and click the red dip to hear the moment the customer got angry.
MVP checklist
- Speaker-labelled transcript
- Sentiment per turn
- Turning-point highlights
- Coaching tips
Tech stack
| Tech | Role | |
|---|---|---|
| Whisper | Speech-to-text | AI |
| LLM | Language model | AI |
| Charts | Data visualisation | UI |
| Next.js | Frontend + API routes | Web |
Reference repos
Open-source projects you can build on. Links checked 2026-10-31.
Related ideas
More AISpeak a two-minute journal entry and see mood and themes over weeks
Beginner10h1–2WhisperLLMCharts16/20Upload a CSV, ask a question and get a chart plus the SQL that produced it
Intermediate14h1–2DuckDBLLMCharts18/20