Dictation Software for Developers and Engineers
Dictate commit messages, docs, and Slack replies without leaving your keyboard. Works in VS Code, GitHub, Linear, and terminal — offline, fast, custom dictionary for code terms.
SuperSpeech is for developers who type 80 words a minute but still find themselves drafting long PR descriptions and design docs. Hit your hotkey, speak the paragraph, and let the cursor do the rest. Unlike cloud dictation that stalls on jargon, SuperSpeech lets you add custom dictionary entries for N8N, kubectl, OAuth, or any other term your cloud tool keeps mangling.
Built for code-adjacent writing
Most dictation tools fail on technical vocabulary because they train on general speech. SuperSpeech lets you add unlimited custom dictionary entries — map "n acht n" to "N8N", "a p i" to "API", "kube control" to "kubectl". Once added, the variant is replaced automatically in every transcription.
- System-wide: works in VS Code, JetBrains, iTerm, GitHub, Linear, Slack
- Custom dictionary for technical terms, acronyms, and CLI commands
- Local processing = works offline on a plane, on a train, in a SCIF
- Paste-in-place output — no clipboard round-trip
- Apple Neural Engine acceleration on M1-M4 Macs
Use cases we see every day
Developers use SuperSpeech for the writing parts of their job that cost the most time:
- Long PR descriptions — "here is why this changes behavior X"
- Design docs and ADRs
- Slack and Linear comments
- Stand-up notes while walking to coffee
- README and docs for internal tools
- Commit messages that are actually complete sentences
Privacy without a VPN
Corporate policy often blocks cloud dictation apps because they upload audio. SuperSpeech runs 100% offline after the initial model download — no network calls, no telemetry beyond optional crash reports. IT approves it the same way they approve any other local tool.
Cross-platform, cross-editor
Mac with Apple Silicon and Windows with NVIDIA/Intel/AMD GPUs are both first-class. Same accuracy, same feature set, same license covers both. Works with any text input — there is no editor plugin to maintain.
Frequently Asked Questions
Does it work in VS Code and JetBrains?
Yes. SuperSpeech dictates system-wide, so it works in any text field including the VS Code editor, JetBrains IDEs, Vim/Neovim GUIs, Sublime, Zed, and iTerm/Warp.
Can I teach it technical vocabulary?
Yes. The custom dictionary maps spoken variants to desired output. Examples: "n acht n" → "N8N", "a p i" → "API", "kube control" → "kubectl". Unlimited entries, case-sensitive toggle per entry.
How fast is it on Apple Silicon?
Under 1 second for 20-30 seconds of audio on M-series Macs. A 1-hour audio file transcribes in about 20-30 seconds offline — roughly 166× realtime.
What about Windows GPUs?
CUDA, TensorRT, and DirectML are all supported. RTX 3060+ gets Mac-class latency. Integrated Intel/AMD graphics run at 1-2 seconds per 30s. CPU-only works in 2-3 seconds.
Does it send any telemetry?
No audio or text is ever sent. Optional crash reports are opt-in. License heartbeats only transmit a hashed hardware ID.
Stop retyping the same PR description
Install SuperSpeech and dictate your next design doc in 3 minutes.