How iOS 18 Spotlight Searches Call Transcripts

With iOS 18, Apple introduces native call recording and automated transcription directly within the Phone and Notes apps. Through on-device intelligence and system-wide indexing, Spotlight can now query the text generated during these phone conversations. This article explains the technical process behind how iOS 18 transcribes phone calls, indexes the resulting text locally, and enables users to retrieve specific spoken moments via a standard Spotlight search query.

1. On-Device Transcription Generation

When a user initiates call recording in iOS 18, the system captures both sides of the audio and uses Apple’s on-device speech-to-text engine to generate a real-time written transcript. Once the call concludes, this transcript—along with the original audio recording—is automatically saved to a dedicated "Call Recordings" folder within the Apple Notes app. Because the conversion from speech to text happens on the Neural Engine, the entire transcript exists as plain, searchable text stored locally on the device.

2. Local Indexing via CoreSpotlight

As soon as the transcript is written to the Notes database, iOS employs the CoreSpotlight framework to index the file. CoreSpotlight parses the transcript's full text, contact names, timestamps, and call metadata. This data is added to the system-level Spotlight index stored in a secure local database. Crucially, this indexing process takes place entirely on the device without uploading call contents or transcripts to external cloud servers, preserving user privacy while enabling rapid search retrieval.

3. Query Matching and Semantic Retrieval

When a user swipes down on the Home Screen or Lock Screen to open Spotlight and enters a query, the search engine compares the input against the local index:

• Lexical Matching: Spotlight performs exact and partial keyword matching against the transcribed words, searching for specific phrases, names, numbers, or topics spoken during the call.
• Semantic Analysis (Apple Intelligence): On compatible devices, Apple Intelligence uses natural language processing to understand the context of the search. This allows Spotlight to identify relevant calls even if the search query uses synonyms or conceptual equivalents of the spoken words rather than an exact quote.

4. Search Results and Deep Linking

If a query matches words spoken within a recorded call, Spotlight displays the result with a Notes or Phone icon, showing a preview snippet with the matched keyword highlighted in context. Tapping the result deep-links the user directly to the transcript inside the Notes app, navigating to the precise timestamp where that specific phrase was spoken so the user can review the text or play back the corresponding audio snippet immediately.