
Memoria
Memoria indexes text, speech, faces and objects in photos and videos locally on the device, allowing users to search their photo albums with keywords, and it does not rely on cloud uploads, accounts or subscriptions.
Updated: 2026-10-27
What is Memoria?
Memoria is a local-first photo and video search app. Its official website states that text, voice and faces can all serve as search entry points. It runs OCR, speech transcription, object recognition and face clustering on the device, turning hard-to-find screenshots, course videos and family photos into searchable content. Compared with cloud album search, its core difference is that data never leaves the phone. The official website emphasizes that indexing is completed on the device, and photos and videos do not need to be uploaded to the cloud first before being searched.
Who is it for?
- iPhone users with a large number of photos and videos who often can't find old content.
- Students and knowledge workers who need to retrieve speeches from course recordings and meeting videos.
- Users who value privacy and do not want to upload family photos to the cloud.
- Individuals who need to quickly organize screenshots, receipts and whiteboard photos.
- Individual users who need to quickly find specific content from a large number of screenshots, receipts and course videos.
Key features
6Photo Text Recognition
Performs OCR on documents, whiteboards, screenshots, and memes, supports multiple languages, and can also recognize text in video frames.
Video Speech Transcription
Uses on-device Whisper to transcribe speech in videos, allowing courses, meetings, and voice notes to be searched by what was said.
Face Clustering
Automatically detect and group human faces, allowing users to find all photos of a person with one click without manual tagging.
Object & Full-text Indexing
Combines object recognition results, transcribed text, and OCR content into a full-text index, allowing you to locate media by entering keywords.
Fully Local Operation
The official website states that there is no need for an account, no cloud uploads, and no tracking. Network requests are only initiated when the optional voice model is downloaded.
One-time Unlock
A free trial is available, and full features are unlocked via a one-time purchase, with no subscription model.
Things to know
Usage Tips
- The initial indexing consumes battery power and time, so it is recommended to complete it when the device is charging and connected to Wi-Fi, especially for albums with a large number of videos.
- First confirm the support for OCR and speech models in the required languages before relying on search results to process important materials.
- Face clustering results may require manual merging or splitting; pay attention to local device access permissions when handling others' photos.
- Although the data is stored locally, it is still recommended to enable system backups and device locks to prevent the loss of irreplaceable data due to a lost phone.
- First, let the album complete a full index, then test different keywords such as people, places, documents, and voices; if a certain type of content cannot be found, first check whether the corresponding language model or permissions are enabled. When organizing the results, group and mark important documents, contracts, and family photos, and regularly confirm that the local backup is available to avoid losing the index and original copies after the device is damaged.
Traffic & popularity
Data month Sep 2026
Memoria receives about 610.6K visits per month.
Monthly visits
5 monthsEstimated from third-party traffic data.
Similar tools
Face++Megvii has launched a computer vision open platform that provides APIs and SDKs such as face detection, comparison, search, liveness detection, portrait processing, optical character recognition, and image recognition. It supports public cloud calls and offline deployment, serving industries such as finance, marketing, and education.
MixpeekMixpeek is a multimodal retrieval platform for agents. According to its official website, it directly reads videos, images, audios and documents in object storage, extracts faces, scenes, voices and screen text, and then locates specific moments with a single hybrid query. It covers scenarios such as material retrieval, advertising creative clustering, content review and quality inspection.
Lenso.aiAn AI-based reverse image search engine, its official website states that it can search billions of images worldwide, presents results classified by people, locations, duplicates, relevance and similarity, and provides research mode, advanced filtering and developer APIs.