Hi Privacy Focused People,

New updates have arrived for Note67

Download at https://note67.com/download

What’s New

v0.1.23 · June 2026

A transcript is only useful if you can actually read it. This release is a
focused pass on exactly that — making what Note67 captures easier to scan, more
accurate, and less cluttered with the noise that speech-to-text tends to leave
behind.

Transcripts that read like prose

Speech recognition naturally chops audio into short, arbitrary chunks. When one
person talked for a while, that used to mean a single wall of text — or a
sentence awkwardly split across several lines mid-thought.

Now Note67 breaks transcripts on sentence boundaries. Each line is a complete
thought, so a long stretch of one person speaking reads like a paragraph instead
of a run-on block. It’s a small change you feel immediately the next time you
skim back through a meeting.

Timestamps that help you navigate

For long monologues, a single timestamp at the top doesn’t tell you much.
v0.1.23 adds interval timestamps every 30 seconds within a speaker’s turn, so
you can jump to roughly the right moment — “what was said around the two-minute
mark” — without scrolling blindly.

Less noise, fewer hallucinations

Local speech models have a habit of inventing things during silence: stray
punctuation, a repeated “Thank you,” or a single word stretched across a long,
quiet gap. We tightened the filtering so that this junk gets dropped before it
ever reaches your transcript, leaving you with the words that were actually
spoken.

Better echo handling

If you listen through speakers instead of headphones, your microphone tends to
pick up the other side of the conversation and transcribe a garbled, duplicated
version of it. v0.1.23 does a better job of recognizing and removing that echo,
so the “you” track stays clean and you’re not reading the same thing twice.

Fixes under the hood

Two reliability fixes round things out:

• Ordering after enhancement — Note67 re-transcribes your audio for accuracy
  once a recording ends. The last line spoken sometimes jumped to the top during
  that step; it now stays put, in the order it was said.
• Search — searching your transcript for terms with special characters (like C++
  or a parenthesis) no longer breaks the view.