Talk it once. Find it forever.
Tap the Voice button in the persistent notification or in the app, talk for up to 60 seconds, and MemX transcribes server-side using OpenAI Whisper. The transcript becomes a searchable memory you can later query in conversation. Voice Query mode lets you ask MemX a question entirely by voice: on-device Silero VAD knows when you stop speaking.
Try MemX FreeHow It Works
Tap Voice, anywhere
From the persistent notification, the lock screen widget, or inside the app. No keyboard, no app launch latency. Just talk.
Whisper transcribes, garbage gets filtered
Audio uploads server-side to OpenAI Whisper. The transcript is filtered for silence and artifacts before it becomes a memory, so your library stays clean.
Search or ask by voice
Find the note later by keyword or by question. Voice Query mode lets you ask MemX in conversation: speak, wait for the auto-stop, get an answer back.
Why Use This Feature
~3× faster than typing on mobile
Speech runs at ~161 words per minute; mobile typing is ~53 WPM. Dictation is roughly 3× faster on a phone (Ruan et al., Stanford / U. Washington, ACM IMWUT, 2017).
Voice is the dominant new input
Juniper Research projected ~8.4 billion voice-assistant devices in use by 2024 (2020 forecast, via Statista). Roughly half of U.S. smartphone users regularly engage with one (eMarketer, 2024). Voice-to-memory ingestion meets people where they already are.
Closes the meeting follow-through gap
Post-meeting action items routinely go unrecorded and uncompleted across industries (Fellow.app State of Meetings; Zoom workplace surveys, 2023-2024). A 30-second voice summary at the end of a call captures the decisions before they fade.
No manual stop button
On-device Silero VAD detects natural speech endings, so you can speak in a normal cadence without holding a button. Useful while driving, cooking, or walking.
Audio you didn’t mean to capture is removed
A silence-and-artifact filter throws out empty seconds and background noise before transcription becomes a memory. Your search results stay relevant.
Becomes a searchable memory like anything else
Transcripts live alongside scanned documents, photo captures, and imported chats. Ask MemX a question and the answer can synthesize across all four sources.