SpeechFast

Product Introduction

1. Definition: SpeechFast is a 100% local, on-device voice dictation and transcription software application for macOS. It is a technical productivity tool that converts spoken audio into clean, formatted text directly at the user's cursor in any application.
2. Core Value Proposition: SpeechFast exists to provide a fast, private, and intelligent dictation experience that works entirely offline. Its core value is enabling users to dictate text anywhere without compromising data privacy, leveraging on-device AI for transcription and Apple Intelligence for text clean-up, eliminating the need for cloud processing.

Main Features

1. 100% Local, Offline-First Architecture: All audio processing occurs on the user's Mac. The speech recognition model (Parakeet or Cohere Transcribe) is downloaded once, and thereafter, no audio data or transcribed text is transmitted over the network. An automated test suite verifies that zero network requests are made during operation, ensuring privacy.
2. Multi-Engine Transcription: Users can choose between three on-device transcription engines. Parakeet (default) is fast, supports 25 languages with auto-detection, and runs on the Apple Neural Engine. Cohere Transcribe offers higher accuracy for critical text, with lower word error rates. Apple utilizes the built-in macOS dictation engine.
3. Intelligent Text Clean-Up: Leveraging Apple Intelligence, SpeechFast automatically removes filler words (e.g., "um," "uh"), false starts, and repetitions from the raw transcript. It also intelligently adds punctuation, resulting in clean, ready-to-use text without manual editing.
4. System-Wide Activation & Simple Gestures: Activated by holding the fn key (or right ⌘) from within any application. A hands-free mode (⇧ + fn) allows for extended dictation. The esc key cancels without pasting. It requires Accessibility permissions to globally detect the hotkey and paste text at the cursor.
5. Local Dictation History: All transcribed text is stored locally in an encrypted history, configurable for retention (default 30 days) or can be turned off. Audio is never stored, only held in memory transiently for processing.

Problems Solved

1. Pain Point: Reliance on cloud-based dictation services that compromise privacy, require a constant internet connection, and often store audio data. SpeechFast solves this by providing a fully offline, private alternative.
2. Target Audience: Professionals working in sensitive environments (legal, healthcare, corporate R&D), frequent travelers (on planes, trains), privacy-conscious users, writers, developers, and anyone who needs to generate text quickly across various apps (Slack, Mail, Notion, code editors) without switching contexts.
3. Use Cases: Dictating confidential emails or reports on a locked-down corporate network; writing code or documentation in a local IDE while offline; composing long-form content in a notes app during a flight; quickly sending messages in communication tools without typing.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros: Unmatched privacy and security (no data leaves the device); works in fully offline environments; intelligent text clean-up reduces editing time; supports multilingual dictation with auto-detection; simple, consistent activation across all apps; one-time purchase license for 3 Macs.
   • Cons: Requires macOS 26 (Sequoia) and an Apple Silicon Mac (M1 or later); dependent on Apple Intelligence for clean-up (must be enabled); initial model downloads are large (600 MB to 2 GB); limited to the accuracy and capabilities of on-device models versus continuously updated cloud services.

2. Key Alternatives & Differentiation:

   • Built-in macOS Dictation: SpeechFast differentiates by adding automatic filler-word removal, punctuation, multi-language auto-detection within a single session, a local history, and a consistent global hotkey. macOS dictation requires manual language switching and outputs raw transcript.
   • Cloud Services (e.g., Google Docs Voice Typing, Otter.ai): These require an internet connection, send audio to remote servers, and often use subscription models. SpeechFast is a one-time purchase with superior privacy and offline functionality.
   • Wispr Flow: A direct competitor offering AI-powered dictation. SpeechFast contrasts with its strict 100% local guarantee, one-time pricing (vs. Wispr's $15/month subscription), and integration of Apple Intelligence for clean-up, versus Wispr's proprietary models.

Frequently Asked Questions (FAQ)

1. How does SpeechFast ensure my voice data is truly private? SpeechFast runs all transcription and clean-up processes locally on your Mac's chip (CPU/Neural Engine). An automated test fails if the app makes any network request during these operations, providing a technical guarantee that no audio or text leaves your machine.
2. Can I use SpeechFast on an Intel Mac or an older version of macOS? No. SpeechFast requires macOS 26 (Sequoia) or later and an Apple Silicon Mac (M1, M2, M3, or later). This is because it relies on the efficiency of Apple Silicon and the on-device Apple Intelligence framework for text clean-up.
3. What happens if I dictate without a text field selected? If no text field is active, SpeechFast will place the cleaned-up transcript directly into your system clipboard and save it to the local history. This ensures no dictation is lost, and you can manually paste it later.
4. How does the language auto-detection work with the Parakeet engine? The Parakeet engine can automatically identify which of its 25 supported European languages you are speaking. You can seamlessly switch languages mid-dictation from one sentence to the next without changing any settings.
5. Is the license a subscription, and how many devices does it cover? SpeechFast is a one-time purchase (€39.99), not a subscription. The license is valid for updates and can be activated on up to 3 of your personal Macs simultaneously. You can deactivate a Mac from the app's settings to free up a slot for a new device.