Call Notes

Why It Exists

Users need to remember important details from phone calls without manual note-taking. Traditional call recording is privacy-invasive and manual note-taking is error-prone.

How It Works

Call Notes captures incoming and outgoing phone calls, transcribes them in real-time using on-device Gemini Nano speech-to-text models, and generates concise AI-powered summaries highlighting key points. All transcription and summarization processing occurs entirely on supported Pixel devices, ensuring conversation privacy. The system can identify speakers, detect action items, and create organized notes that sync with Google Keep and other productivity apps. Call Notes was introduced with the Pixel 9 series and leverages the Tensor G4 NPU for efficient local processing.

Everyday Use Cases

• Business call summaries
• Doctor appointment notes
• Meeting action items
• Interview follow-ups
• Legal consultation notes
• Family conversation reminders

User Workflow

• User enables Call Notes for phone calls
• Audio is captured during the call
• Gemini Nano transcribes speech to text in real-time
• AI generates a summary of key points
• Summary is saved and synced to Google Keep

AI Processing Flow

Call audio is captured by the phone's microphones and processed by the on-device Gemini Nano speech recognition model. Speaker diarization separates different voices. The transcribed text is analyzed by a Gemini Nano summarization model that identifies key topics, action items, and decisions. Summaries are organized by speaker and time. All processing happens on the Tensor G4 NPU with no data leaving the device.

Inputs / Outputs

Inputs:

• Call audio streams
• Speaker identification data
• Timestamp metadata
• Contact information
• Call type metadata

Outputs:

• Full transcription
• Action item list
• Key decision summary
• Speaker-separated notes
• Synced Google Keep entries
• Searchable call history

Known Limitations

• Limited to supported languages at launch
• Requires Google account for sync features
• Accuracy may vary with background noise
• Available only on Pixel 9 series and newer
• Some accents may reduce transcription quality

Unsupported Scenarios

• Conference calls with 10+ participants
• Calls in extremely noisy environments
• Real-time translation during calls
• Voicemail transcription
• International roaming calls

Performance Notes

Real-time transcription adds less than 100ms latency; Summary generation completes in under 5 seconds after call ends; Entire transcription available immediately after call; NPU power consumption under 3 percent per hour of call

Available On

Shows where this feature is available and how its AI processing works on each platform. Availability may vary by device.

Research Status

Confidence and verification reflect how complete documentation is. Fields may show Not assessed or Not yet verified while research is ongoing - this flags gaps, not product deficiencies.

Research Notes

Available on Pixel 9 series from launch in August 2024; Requires Google Tensor G4 or later; Uses Gemini Nano 2 speech model; Introduced at Google I/O 2024; Synced to Google Keep and Gmail; End-to-end encryption supported

Details

Capability Mapping

This feature maps to canonical capability family: Translation.

Canonical capability: Message Translation.

• Message Translation - Translates text messages between languages.

Grounded in 8 device(s) on Gemini Nano.

Related Features

• AI Call Assistant - Cross-platform equivalent