Comparing DictateKeyboard gainst WisprFlow

[yaroslavvb]

I noticed today that your service is getting quite close to WisprFlow which I'm paying money for. I would get rid of them if you could make it equally good.

WisprFlow had this regression where they sometimes drop the first 2 seconds of speech, and they've had this regression for 4 months. So that's quite annoying. On the other hand, they seem to be much faster.

If there is a way to get similar speed, using paid Soniox or anything else, I would be a super loyal user.

Here's what the current experience looks like for me, side by side

screen-20260916-141210-1789593113614.mp4

[DevEmperor]

Hey, I personally used Groq as a provider for a long time because it’s very fast, very accurate, and completely free for normal use. A while ago, I switched to Soniox because it’s also extremely affordable for personal use. At the same time, it supports real-time processing and is even more accurate. In terms of speed, it’s really hard to beat, and I can highly recommend it. :)

[yaroslavvb]

I'm using Soniox as well, but it was slow. I just realized by looking int Transcription History that I'm using the async model. I spent the last two minutes clicking around the menu, and I can't figure out how to enable the streaming one.

[yaroslavvb]

I'm on version 6.2.0 (140)

[DevEmperor]

Okay, I don't know how Soniox behaves in async mode. For realtime, just go to the Dictate Settings -> Dictate -> Recording -> Real-time transcription. There you should select the second item. Then Dictate should automatically switch to Soniox Real-time model (stt-rt-v5). :)

[yaroslavvb]

nice, I think maybe that should be the default, async version is not meant to be real-time, so latencies are meant to be unpredictable. (They are discussing making a fast sync version, but the current one isn't)

[yaroslavvb]

Holy cow it's so fast!!!