Revi

Product Introduction

1. Definition: Revi is a cross-platform (macOS, Windows, Linux) on-device voice dictation and clipboard management utility software. It functions as a privacy-focused speech-to-text (STT) engine, operating entirely locally without cloud dependencies.
2. Core Value Proposition: Revi provides fast, private, and offline voice-to-text dictation and advanced clipboard management for desktop computers, using free open-source speech recognition models. It solves the core problem of sending sensitive voice data and keystrokes to third-party cloud servers by keeping all processing on the user's local machine.

Main Features

1. Instant On-Device Dictation: Revi enables real-time speech recognition across any text field on the operating system. How it works: The user holds the ⌥ (Option/Alt) key, speaks, and releases the key. Text is transcribed and inserted directly at the cursor location with sub-half-second latency. This "hold and speak" mechanic eliminates the need for dedicated dictation windows or interfaces.
2. Multi-Engine & Multilingual Processing: The application offers a choice of on-device speech recognition engines. Users can select between Apple's native dictation engine (macOS), NVIDIA's open-source Parakeet model, or the full catalog of OpenAI Whisper models (from Tiny to Large v3). It supports automatic detection and transcription for over 25 languages without manual configuration.
3. Intelligent Local Text Cleanup: Post-transcription processing includes automatic punctuation, capitalization, and removal of verbal fillers like "uhm." This cleanup is performed locally on-device, leveraging Apple Intelligence on macOS or standard processing on other platforms, with a strict rule against paraphrasing content to ensure user text integrity.
4. Privacy-Conscious Design: The software is engineered with a zero-telemetry, zero-account philosophy. It explicitly avoids typing text in password fields, ensuring credentials remain secure. No user data, audio, or text ever leaves the computer.
5. Advanced Clipboard Manager: Revi integrates a persistent clipboard history manager that records up to 10,000 copied items, including text and screenshots. It is accessible via a keyboard shortcut (⇧⌘V), allowing users to filter and paste historical items, significantly enhancing productivity for copy-paste-heavy workflows.
6. Minimal Footprint & Offline Operation: The core application is approximately 12 MB, with zero idle CPU/GPU usage. It functions completely offline, making it suitable for use on airplanes, in secure facilities, or in locations with no internet access.

Problems Solved

1. Privacy & Data Security Risks: Addresses the critical concern of sending voice audio and screen context to cloud servers for processing, which is a standard practice for many dictation services. Revi eliminates this data exposure risk.
2. Subscription Fatigue & Opaque Pricing: Provides a solution to the recurring monthly subscription fees (e.g., ~$15/month) common in the dictation software market. Revi's one-time payment model offers lifetime access to core features.
3. Offline Inaccessibility: Solves the problem where popular dictation tools become non-functional without an internet connection. Revi's fully local engine ensures reliability in any environment.
4. Inefficient Text Input & Clipboard Management: Addresses slow typing speeds and the poor native clipboard history limitations in most operating systems. The integrated dictation and powerful clipboard manager streamline both input and data retrieval workflows.
5. Target Audience: Privacy-conscious professionals, writers, developers, students, writers, data entry specialists, and any desktop user seeking to increase text input efficiency without compromising data security.
6. Use Cases: Composing emails and documents in secure environments, taking meeting notes offline, quick note-taking while driving (via hands-free mode), managing extensive code snippets or design assets with the clipboard manager, and multilingual content creation.

Unique Advantages

1. Differentiation: Unlike cloud-dependent dictation apps (e.g., many versions of Dragon, Google Voice Typing, or subscription services like Wispr Flow), Revi processes all data locally. Its primary differentiator is its strict "no cloud, no account, no subscription" model for core features, combined with a competitive one-time fee (€7.99).
2. Key Innovation: The innovation lies in packaging and optimizing free, open-source STT models (Whisper, Parakeet) into a user-friendly, cross-platform desktop application with a seamless "hold-key-to-dictate" UX. It also innovates by deeply integrating a powerful clipboard manager focused on security (e.g., not recording passwords from managers) and productivity.

Frequently Asked Questions (FAQ)

1. Does Revi send my voice recordings or typed text to the cloud? No. All speech recognition, text cleanup, and processing for the core dictation and clipboard features occur entirely on your local computer. Your voice never leaves your device, and no account is required for use.

2. How does Revi compare to cloud dictation services in terms of accuracy? For local processing, Revi's accuracy depends on the chosen model and your hardware. The larger Whisper models (e.g., Large v3) offer accuracy comparable to cloud services, but require more RAM (16GB recommended). For absolute maximum accuracy, Revi offers an optional, clearly-labeled Cloud plan (coming soon) using OpenAI's API with enhanced features like mid-sentence language switching.

3. What is the difference between the €7.99 one-time purchase and the future Cloud plan? The €7.99 purchase is for the application license, granting lifetime access to all on-device dictation engines and features. The optional Cloud plan (a future subscription) will provide access to more powerful cloud-based processing (OpenAI Whisper) and advanced rewriting capabilities, but the core local functionality remains permanent with the one-time purchase.

4. Is Revi compatible with my operating system and hardware? Revi supports macOS (Apple Silicon), Windows 10/11, and Linux, all in 64-bit. While basic engines like Parakeet run on most hardware, the largest Whisper models for top accuracy benefit from at least 16GB of system memory.