Made by one person who got tired of typing.
Speechcap is built by Dhruwang Jariwala, operating as Speechcap Labs out of Surat, India.
Why this app exists
I write a lot — code comments, Slack messages, PR descriptions, Notion docs, emails to investors, essays — and at some point I noticed I was typing thoughts I'd already had in my head. The bottleneck wasn't my brain. It was the keyboard.
I tried Wispr Flow, Superwhisper, Apple Dictation, and a few others. They all worked. Two things bothered me:
- Most of them sent my audio to a cloud somewhere. For a tool I'd be talking into all day, that felt wrong even when the content wasn't sensitive.
- The subscription pricing was disproportionate to what the tools did. $15/month for “voice in" when an iPhone subscription is the same money is hard to justify.
So I built Speechcap. Both stages — transcription via Whisper and AI cleanup via a small local LLM — run entirely on your Mac on Pro. The price is half of what the closest alternatives charge, with PPP adjustment so it's genuinely affordable globally.
What I'm optimising for
- Privacy as the default, not the upgrade. Even the Free tier transmits audio only transiently and never retains it.
- Push-to-talk over tap-to-toggle.If the mic is on, your finger is on the key. There's no ambiguity to misremember.
- Real local-first AI. Both the speech model and the cleanup model run on your hardware. The Mac you already own is more than enough.
- Price that respects where you are. $6 in San Francisco, $3 in Bengaluru, comparable prices everywhere else. Not the US sticker for everyone.
What I'm not doing
Not chasing meetings (that's Otter's job), not building voice assistants (that's every other AI company's job), not adding features that'd compromise the privacy claim. The product is intentionally narrow.
Talk to me
Email is the fastest way: hello@speechcap.com. I read everything personally. Bug reports, feature ideas, “your hero copy is unhinged” — all welcome.