[Suggestion] Streamlit real-time audio transcription dashboard with Audio Intelligence (Python)

[deepgram-robot]

What to build

A Streamlit web application that provides a real-time audio transcription dashboard — users can upload audio files or record from their microphone, get transcriptions with Nova-3, and see Audio Intelligence results (sentiment analysis, topic detection, entity extraction, summarization) visualized in interactive charts and tables.

Why this matters

Streamlit is the most popular framework for building data science and ML demo apps in Python, with millions of monthly users. Data scientists and ML engineers evaluating speech APIs often prototype in Streamlit first. Having a ready-made Streamlit dashboard that showcases Deepgram's full feature set (transcription + Audio Intelligence) gives these developers an instant starting point and demonstrates capabilities they can show to stakeholders without building a custom frontend.

Suggested scope

• Language: Python
• Framework: Streamlit
• Deepgram APIs: STT (Nova-3 pre-recorded), Audio Intelligence (sentiment, topics, entities, summarization, PII detection)
• Features:
  • File upload for audio transcription
  • Microphone recording via st.audio_input
  • Transcription display with speaker diarization
  • Sentiment timeline chart (plotly/altair)
  • Topic distribution pie chart
  • Entity extraction table
  • Summary panel
  • PII detection highlights
• Complexity: Intermediate
• Backend: Python only (Streamlit handles the UI)

Acceptance criteria

• Runnable with minimal setup (pip install -r requirements.txt, add API key, streamlit run)
• README explains the dashboard features with screenshots
• Uses current SDK version
• Supports both file upload and microphone recording
• Visualizes at least 3 Audio Intelligence features with charts/tables
• Responsive layout that works on desktop browsers

---

Raised by the DX intelligence system.