Show HN: RealtimeVAD – trading false positives for 110ms earlier end-of-speech
discuss
- Points26
- Comments5
discuss
Hey guys. I recently watched the latency reports for my local voice assistant, which show a waveform plus silero probability for voice activity. I noticed silero lags visibly on the speech off signal. There had to be a method to get this signal closer to what I felt is optimal. My idea was to use a 10 ms energy detector with a threshold that adapts to environment or room noise. Which worked amazingly well. It created more false positives but in my speculative generation pipeline that would not…
Story details from Hacker News ↗ · linked article on github.com.
Show HN: RealtimeVAD – trading false positives for 110ms earlier end-of-speech was first seen on Hacker News · Front page on Nov 12, 2026. Since then it has appeared in 73 checks on 2 different days, across 2 lists (Hacker News · Front page, Hacker News · Show HN). Its best position was #16 of 30, reached on Nov 13, 05:13 UTC. It dropped off the list after Nov 13, 21:22 UTC.
Position on Hacker News · Front page at every check (lower is better).
Best rank per day (UTC).
Platform metric as reported by Hacker News.
| List | Best | Days | Checks | Last seen |
|---|---|---|---|---|
| Hacker News · Front page | #16 | 1 | 27 | |
| Hacker News · Show HN | #1 | 2 | 46 |
A few top-level comments from Hacker News, shortened. Read the full thread on Hacker News.
You can also avoid the problem all together. I made a "speculative regeneration" at least thats what i call it pipeline i been using for many months now and it works great. basically as long as you have access to the inference server you are good to go. idea is as such, every n milliseconds of silence (usually 300ms)…
This is the 3rd or so time I've posted this - Run a tiny shitty llm, one that is super fast. Or better yet try out a decision model now days. Feed it the previous conversation turn and all the STT (ASR output) you have so far and just ask it for a probability that the user is done speaking. Use that probability to…
Archived daily, weekly and monthly lists that include this entry.
Not at the moment. It was last on the list on Nov 13, 21:22 UTC, after 2 days of trending.
#16 on Hacker News · Front page, on Nov 13, 05:13 UTC. Trendoday checks this list every 15 minutes.
It was first seen on Nov 12, 2026 and has been on the list on 2 days (73 checks) since.
26 points and 5 comments; submitted by koljab on Nov 12, 2026.