NotelyVoice: Open Source Note Taking & Knowledge Management

NotelyVoice

“A 100% private AI voice transcription app that converts speech to text in 100+ languages. Built with Compose Multiplatform for Android & iOS using Whisper AI - ”

A 100% private AI voice transcription app that converts speech to text in 100+ languages. Built with Compose Multiplatform for Android & iOS using Whisper AI - no cloud uploads, all processing happens on-device for complete privacy.

androidaudio-playercompose-ioscompose-multiplatformios jetpack-composekoin-mvvmkotlin-coroutines

maintainer? embed the star badge

get the markdown →

What is NotelyVoice?

NotelyVoice is a free, open-source, cross-platform voice transcription and note-taking application for Android and iOS that converts speech to text in more than 100 languages entirely on-device using OpenAI Whisper, built for students capturing lectures, professionals documenting meetings, doctors recording patient notes, and researchers transcribing interviews.

What it is

NotelyVoice is a mobile note-taking application that combines rich text editing, audio recording, playback and speech recognition in a single app. It lives in the Kotlin Multiplatform and Compose Multiplatform ecosystem: the repository badges record Kotlin 2.1.21, Compose 1.8.0 and Android API 21+ support, and its topic list describes a Koin MVVM structure with Kotlin coroutines, with the same codebase targeting both Android and iOS. Two distributions exist. This repository is the open-source version, released freely on F-Droid under GPL-3.0 and actively maintained; a separate version with a rebuilt codebase, a redesigned interface and a subscription model is published on Google Play, and revenue from that build funds development of both.

The concrete problem it solves is that speech recognition normally depends on a network round trip, which means private recordings leave the device before they become text. NotelyVoice performs that processing locally, so speech recognition works without an internet connection and no cloud uploads take place. It also removes the need to move between a recorder, a transcription service and a separate notes app: dictation goes directly into a note, and recorded voice notes can be played back and transcribed in place with unlimited transcriptions into multiple languages.

Key capabilities

On-device speech-to-text powered by OpenAI Whisper, which can run locally without an internet dependency once the model is downloaded.
Rich text editing with headers and sub-headers, title styling, bold, italic and underline, plus left, right and centre text alignment.
In-app audio recording and playback, with recorded audio and note text shareable to destinations such as Messages, WhatsApp, Files and Google Drive.
A streaming WAV decoder that processes audio in chunks instead of loading whole files into memory, avoiding out-of-memory errors on long recordings.
Overlapping chunk transcription with a default chunk size of 30 seconds, three seconds of overlap, a configurable maximum chunk size of 60 seconds and a 1 MB processing buffer.
Note organisation through folders and tags, text search, and filtering by Starred, Voice Notes or Recent.
Light and dark theming with a consistent experience across Android and iOS.

Who uses it and how

Students recording lectures and dictating study notes, including in places where a reliable connection is unavailable and offline recognition is the only workable option.
Professionals documenting meetings, either by dictating straight into a note or by recording audio and transcribing it afterwards.
Clinicians capturing patient notes, where keeping audio on the device rather than uploading it to a service is the point of the tool.
Researchers transcribing interviews, including recordings long enough that the chunked, memory-efficient pipeline is needed to avoid out-of-memory failures.
Individuals who want hands-free note-taking shared across an Android phone and an iOS device, since both are produced from this repository.

Getting started

The open-source build is distributed freely on F-Droid under GPL-3.0, with the project homepage at notelyvoice.com, and the source compiles from this repository with Kotlin 2.1.21 and Compose 1.8.0 for Android API 21+ and iOS.

How it compares

The closest named technology is OpenAI Whisper, the open-source speech recognition model that NotelyVoice embeds rather than reimplements, supplying the mobile interface, recording, note editing and chunked inference pipeline around it. The other point of comparison is its own sibling build on Google Play, which is proprietary, rebuilt from scratch with a redesigned UI and sold by subscription, whereas this repository is GPL-3.0 and free.

When to use it — and when not to

Because this is an on-device mobile app rather than a server, a self-hoster does not operate a database, object storage or SMTP service; the burden is building and shipping the Android and iOS targets and managing model downloads. Anyone who wants the redesigned interface, a subscription-backed release channel or vendor support should choose the Google Play version instead. It is also worth checking expectations against the project itself: the README's tagline claims 100+ languages while its speech recognition section cites Whisper support for over 50, the repository carries 44 open issues, and the last recorded push is 21 August 2026.

project readme (upstream, from github) — read inline

Notely Voice

A completely free, modern, cross-platform, 100% private AI voice transcription & note-taking application with powerful Whisper AI Voice to Text capabilities built with Compose Multiplatform.

Perfect for students capturing lectures, professionals documenting meetings, doctors recording patient notes, researchers transcribing interviews, and anyone needing accessible hands-free note-taking across all their devices.

Two versions of NotelyVoice This repository is the open source version, distributed freely on F-Droid under GPL-3.0 and actively maintained. A separate, fully rebuilt version with a new codebase, redesigned UI, and a subscription model is available on Google Play. Revenue from the Play Store version funds ongoing development of both versions.

	F-Droid (this repo)	Google Play
Price	Free	Subscription
Source	Open source (GPL-3.0)	Proprietary
Codebase	This repository	Rebuilt from scratch
UI	Original	Fully redesigned
Updates	Active	Active

Download the app

Screenshots

screenshot2 screenshot3

Features

Note-Taking

✏️ Rich Text Editing - Format your notes with:

Headers and sub-headers
Title styling
Bold, italic, and underline text
Text alignment (left, right, center)

🔍 Simple Search - Find your notes instantly with text search
📊 Smart Filtering - Filter notes by type (Starred, Voice Notes, Recent)
📂 Organization - Categorize notes with folders and tags

Speech Recognition

🎙️ Advanced Speech-to-Text - Convert speech to text with high accuracy
🌐 Offline Capability - Speech recognition works without an internet connection
🔄 Seamless Integration - Dictate directly into notes or transcribe audio recordings
🎧 Audio Recording - Record voice notes and play them back within the app
🎧 Unlimited Transcriptions - Transcribe unlimited voice notes to multiple languages

General

🌓 Theming - Switch between dark and light themes based on your preference
💻 Cross-Platform - Seamless experience across Android & iOS
📱 Share Audio Functionality - Share audios recorded on the App to Messages, WhatsApp, Files, Google Drive etc
📱 Share Texts - Share texts on the App to Messages, WhatsApp, Files, Google Drive etc

Audio Processing Improvements

Memory-Efficient Audio Processing

The app now includes memory-optimized audio processing to handle large files without Out of Memory (OOM) errors:

Streaming WAV Decoder: Processes audio files in chunks instead of loading entire files into memory
Overlapping Chunk Transcription: Splits long audio files into manageable chunks with overlap for seamless transcription
Configurable Chunk Sizes: Adjustable chunk sizes based on device memory and performance requirements

Chunking Configuration

The transcription process uses configurable parameters:

Default Chunk Size: 30 seconds per chunk
Overlap Size: 3 seconds overlap between chunks
Maximum Chunk Size: 60 seconds (configurable)
Memory Buffer: 1MB processing chunks

Benefits

No More OOM Errors: Large audio files (hours long) can be processed without memory issues
Better Performance: Smaller chunks process faster and use less memory
Seamless Transcription: Overlapping chunks ensure no text is lost at chunk boundaries
Progressive Processing: Real-time progress updates as chunks are processed

Speech Recognition Technology

OpenAI Whisper - State-of-the-art open-source automatic speech recognition
- Robust multilingual speech recognition with support for over 50 languages
- Trained on 680,000 hours of multilingual and multitask supervised data
- Excellent performance across diverse audio conditions and accents
- Can run locally without internet dependency once model is downloaded
Cross-Platform Compatibility - Designed for versatile deployment
- Consistent speech recognition quality across different operating systems
- Flexible model sizes from tiny (39 MB) to large (1550 MB) based on accuracy needs
- Advanced noise robustness and speaker independence
- Perfect for applications requiring high-quality transcription in various environments

Built With 🛠

Kotlin - Official programming language for Android development
Compose Multiplatform - UI toolkit for building native applications
Coroutines - For asynchronous programming
Android Architecture - Ensures scalability and testability
ViewModel - Stores and manages UI-related data
Koin - Dependency injection for Android
Material 3 - Design system for modern UI
Whisper AI - Human level robustness speech recognition
Native Compose Navigation - No third-party navigation libraries
Custom Text Editor - Built from scratch without external editing libraries

Architecture

Notely is built with Android Architecture principles, separating the app into distinct layers:

UI Layer: Compose UI components
Presentation Layer: Platform Independent ViewModels
Domain Layer: Business logic and use cases
Data Layer: Repositories and data sources

Project Structure

shared/: Contains shared business logic and UI code.

androidApp/: Contains Android-specific code.

iosApp/: Contains iOS-specific code.

Contributing

Contributions are welcome! Please follow these steps:

Fork the repository.
Create a new branch for your feature or bug fix.
Submit a pull request with a clear description of your changes.

Getting Started

Prerequisites

Android Studio Ladybug or newer
XCode 16.1
JDK 11 or higher
Kotlin 2.0.21 or higher

Installation

Clone the repository

git clone https://github.com/tosinonikute/NotelyVoice.git

Open the project in Android Studio
Sync the project with Gradle files
Run the app on an emulator or physical device

License

Copyright (C) 2025 NotelyVoice

This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License version 3 only 
as published by the Free Software Foundation.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program. If not, see <https://www.gnu.org/licenses/>.

Frequently asked questions

Is NotelyVoice free to use?

NotelyVoice is open source under the GPL-3.0 licence. There is no licence fee and no seat count — you can self-host it or, where the project offers one, pay a vendor for a managed version instead.

What does NotelyVoice do?

A 100% private AI voice transcription app that converts speech to text in 100+ languages. Built with Compose Multiplatform for Android & iOS using Whisper AI -

What is NotelyVoice written in?

NotelyVoice is primarily written in C++. Its source is publicly available at https://github.com/Notely-Voice/NotelyVoice, and it has 793 GitHub stars.