SpeakoFlow

Product Introduction

1. Definition: SpeakoFlow is a free, open-source (MIT-licensed) desktop application that functions as a comprehensive voice-to-text and AI-powered productivity assistant. It operates as a system-wide utility on Windows, macOS, and Linux, enabling voice interaction across any application.
2. Core Value Proposition: It exists to eliminate the friction between thought and digital execution by replacing manual typing with intelligent voice commands. Its primary value is enabling hands-free, context-aware computing where users can dictate, generate, translate, and command an AI assistant using natural speech, all while prioritizing user privacy with on-device speech-to-text processing.

Main Features

1. Universal Dictation: This feature provides real-time, low-latency speech-to-text transcription directly into any active text field. It uses a local, on-device transcription model (based on the Handy project's core) to convert speech to text without cloud dependency. Users activate it with a global hotkey (default: Left Ctrl + Left Super), speak, and words appear live or in a batch.
2. Flow Command ("Hey Flow"): An AI text generation trigger integrated into the dictation stream. When a user prefaces their speech with the "Hey Flow" trigger (customizable), the system interprets the following speech as a command (e.g., "reply to this email," "draft a prompt"). It uses the configured AI model (local or cloud) to generate the requested content and automatically pastes the finished text into the active application.
3. Screen-Aware AI Assistant: A floating, interactive AI panel summoned by a hotkey (Left Ctrl + Left Alt). This assistant can perform two key actions: answering general queries and, crucially, analyzing screen content. When permitted, it uses OCR and vision models to understand the context of the user's active window (error messages, charts, emails) to provide relevant, contextual answers. It supports streaming text responses and text-to-speech audio replies.
4. AI-Powered Text Cleanup & Translation: Post-dictation processing features. AI Cleanup takes raw, disfluent dictation (with "ums," false starts) and applies grammatical correction, filler-word removal, and tone adjustment (Professional, Friendly, Concise, or custom). Live Translation allows users to speak in one language (e.g., Spanish, Hindi, Japanese) and have clean English text typed directly into their application.
5. Local Memory & User Profiles: An optional, privacy-focused personalization system. When enabled, it creates a local, editable memory of user preferences and writing style to help the assistant generate more personalized responses. Profiles allow users to switch between different AI personas (e.g., "warm companion," "quick-answer mode") with preset tones and response lengths for different tasks.

Problems Solved

1. Pain Point: The inefficiency and physical strain of constant typing, especially for long-form writing, coding, or communication. It also addresses the fragmentation of having separate tools for dictation, AI assistance, and translation.
2. Target Audience: Content Creators & Writers (bloggers, authors, journalists), Developers & Engineers (documenting code, writing terminal commands), Knowledge Workers & Managers (handling high-volume email, chat, and report writing), Accessibility Users (individuals with RSI, carpal tunnel, or mobility impairments), and Multilingual Professionals who think or communicate in multiple languages.
3. Use Cases: Hands-Free Email Composition: Dictating and generating full email replies while reviewing the incoming message on screen. Code Documentation & Commenting: Using voice to write inline comments or documentation without breaking coding flow. Live Meeting Note Translation: Speaking notes in a native language during a meeting and having them transcribed in English in real-time. Debugging Assistance: Showing a code error or terminal output to the screen-aware assistant for an instant explanation.

Unique Advantages

1. Differentiation: Unlike closed-source, subscription-based competitors (Wispr Flow, Aqua Voice, Typeless), SpeakoFlow is completely free and open-source (MIT license). It offers a more integrated feature set than basic dictation tools by combining local transcription, an AI command layer ("Hey Flow"), a contextual assistant, and translation into a single package. It provides more privacy than cloud-only tools by keeping speech-to-text entirely on-device.
2. Key Innovation: Its modular and private architecture is its core innovation. It decouples the mandatory private component (speech-to-text) from the optional AI components, allowing users to choose their model provider (fully local, self-hosted server, or any cloud API with their own key). The screen-aware context engine that can be manually triggered, set to auto-decide, or disabled offers granular control over privacy and functionality not commonly seen in consumer assistants.

Frequently Asked Questions (FAQ)

1. Is SpeakoFlow really free and how does it compare to paid alternatives like Wispr Flow? Yes, SpeakoFlow is completely free and open-source under the MIT license, with no account or subscription required. Unlike Wispr Flow (closed-source, cloud-based, with weekly word limits), SpeakoFlow offers on-device transcription, a more comprehensive feature set including the "Hey Flow" generator and screen-aware assistant, and full control over your AI model and data, all at no cost.
2. How does SpeakoFlow handle privacy and is my voice data sent to the cloud? SpeakoFlow is private by default. The core speech-to-text transcription runs locally on your machine using an offline model; your voice audio is never uploaded. For AI features (like "Hey Flow" or the assistant), you control the provider. You can use a fully local model, your own server, or a cloud API with your own key, meaning SpeakoFlow itself does not collect or store your data.
3. What is the difference between the "Hey Flow" command and the AI assistant panel? "Hey Flow" is a voice command used during dictation to generate and paste finished text (like an email reply) directly into your current app. The AI assistant is a separate, persistent chat panel you summon for back-and-forth conversation, screen analysis, and spoken answers without leaving your primary application. They are two different interfaces to the same underlying AI model.
4. Can I use SpeakoFlow completely offline without any internet connection? Yes, you can run SpeakoFlow in a fully offline mode. The transcription engine works offline. By using the built-in local AI model (or another locally hosted model like Ollama) for the "Hey Flow" and assistant features, the entire application can function with no internet connectivity.
5. Why does macOS say SpeakoFlow "can't be opened" and how do I fix it? This is a common macOS Gatekeeper warning for unsigned applications from developers without a paid Apple certificate. It does not mean the software is harmful. To open it, drag the .app to your Applications folder, open Terminal, and run the command xattr -dr com.apple.quarantine /Applications/SpeakoFlow.app once. You only need to do this once per new version you install.