Time to work on Emacs News!

Raw material: https://sachachua.com/blog/2026/11/2026-11-09-emacs-news/

We have a recording session for https://emacsconf.org/2026/talks/community/ tomorrow, so I need to get my thoughts together. Also, if there's time, I want to to document and demonstrate my video workflow so that I can do it again next year, and maybe look into organizing my windows and workspaces in preparation for trying to manage everything everywhere all at once. Aaah! Totally not panicking.

We have a recording session for https://emacsconf.org/2026/talks/community/ tomorrow, so I need to get my thoughts together. Also, if there's time, I want to to document and demonstrate my video workflow so that I can do it again next year, and maybe look into organizing my windows and workspaces in preparation for trying to manage everything everywhere all at once. Aaah! Totally not panicking.

EmacsConf has two days of two tracks, and my brain is not very good at paying attention to things that have to be done at a certain time. So I use cron to schedule the playback of talks at the scheduled times during the conference, including:

• announcing the talk on IRC
• playing the recorded intro, if it exists
• playing the video file or joining the talk's BigBlueButton room
• going to the Q&A session, which could be in BigBlueButton, Etherpad, or IRC
• publishing the talk video files on the wiki and on media.emacsconf.org

This came in really handy last year because I had to manage EmacsConf by myself for part of the conference. This year, I think I might need to do the same.

One of the challenges of automating something that I do only once a year is that I tend to forget about things, especially if I quickly put together the automation in a rush because I was trying to get things done. If I didn't take the time to document them while it was still fresh in my brain (because I needed to move on to doing something else), now I need to dust off the code, figure out how it works, and fill in those missing notes.

Today I want to get that automatic schedule working again. If I have time, maybe I will record the introductions.

orga - user that the /usr/local/bin/handle-session script connects to

I can demonstrate my workflow for extracting the clips from my recording of a pronunciation practice session with my teacher.

workflows#extract-my-french-pronunciation-lessons-tongue-twisters-virelangues

I'm about halfway through editing the transcript. I would like to get it nicely edited so that it'll be easy to pull out snippets of it as text.

Draft: https://sachachua.com/blog/2026/11/my-chat-with-linkarzu/

Time to work on Emacs News!

Raw material: https://sachachua.com/blog/2026/11/2026-11-02-emacs-news/

I finally get to work on this, at least until the kiddo wakes up! I want to turn this into the usual nicely searchable transcript. Draft: https://sachachua.com/blog/2026/11/my-chat-with-linkarzu/

Possible things to show

process of drafting a blog post

• Why:
  • I like writing blog posts because it's a lot easier for people to skim than a twenty-minute video.
  • People can copy code from it, I can search it, um, I can update it as I learn more or as I fiddle with the code.
  • Text is so much easier to work with.
• Challenges:
  • Some people think very logically.
    • They can start with a top-down outline and then systematically fill it in.
  • Some people work bottom up.
    • They collect a lot of notes which they organize them into larger and larger chunks, and they they make something useful that way.
  • My mind tends to wander all over the place.
    • Sometimes I work from the top down.
    • Sometimes I want to jump to a different part of my outline to fill something in.
    • Sometimes I just have some sentences and then I need to figure out how to get from one place to another.
• How
  • I wrote about my writing workflow in 2025 (Finding the shape of my thoughts).
  • My workflow has changed a little bit since then.
    • I spend much less time waiting around during playdates these days; it's getting colder because the kiddo now mostly prefers to spend time with me or playing Minecraft. Instead of using my phone for outlines, snippets, placeholders, Uh, she's, you know, maybe she and her friends are growing a little bit apart. Middle school is hard, etc., etcetera.
    • Um
    • So most of my writing now is done on my laptop and the audio br brain dump then doesn't come as a monolithic audio file or this big wall of text that I then have to edit and sort through.
    • It by using this.
    • This workflow.
    • I can.
    • Um, use the outline as a visual anchor for my thoughts while using dictation to quickly jot things down.
    • And all of that is very useful and I should have remembered to
  • Publishing.
  • Next steps.
  • If I'm untangling my thoughts about a complex topic, sometimes I start with a sketch to figure out what I want to say and the order in which I want to say them.
  • Normally, when I'm writing, I do like to play around with sometimes putting one sentence per line.
    • This allows me to move the sentences around very quickly using Org shortcuts, and it also helps me see the variety in the lengths of sentences.
    • I do tend to have very long sentences.
    • It reminds me to break them up into shorter ones.
  • getting the captions for the current recording

Since I can use Emacs and a bit of Python to visualize the gaps and the conversation balance in real-ish time, and I can use speech recognition to see if I've said any of my target words or phrases, why not use that speech recognition result to make a proper VTT of the current recording on the fly so that I can jump to things or extract stuff?

I generated a modified asr.py that gives me Unix timestamp * 1000 + milliseconds. I can check the timestamps against the timestamp in the filename of the latest recording, which OBS is still adding to. This lets me approximately synchronize things so that I can jump to parts of the current recording while I'm recording it, and I can adjust the timestamps further by referring to the waveforms.

Respond to irreal, build on the idea some more

rfc822msgid:20050329231145.62086.mail@mail.emailprovider.com

Time to work on Emacs News!

Draft with the raw links: 2026-10-26 Emacs news

Yesterday, I added voice in and voice out to M-x doctor. I think I've figured out how to get gptel to do the same so that I can practice talking in French with Emacs. There are a number of mobile applications that offer similar kinds of AI voice conversations for a subscription, but I'd prefer to do things in Emacs, of course, so that I can use the data and tweak things more. I'll demonstrate what I've got working so far and then tinker with it to see what else I can come up with.

Outline:

• M-x doctor and sacha-whisper-conversation-mode
• sacha-gptel-practice-french
• https://github.com/jwiegley/pending would be nice
  • speech recognition

  • gptel chat

    (use-package pending
      :vc (:url "https://github.com/jwiegley/pending"))
    
    (defvar sacha-whisper-pending-overlay nil "Overlay that will have the pending indicator.")
    (defvar sacha-gptel-pending-overlay nil "Overlay that will have the pending indicator.")
    
    ;;;###autoload
    (defun sacha-whisper-run-with-pending-indicator ()
      "Display a pending indicator while waiting."
      (interactive)
      (if (process-live-p whisper--recording-process)
          (pending-update sacha-whisper-pending-overlay
                          :label "💭")
        (setq sacha-whisper-pending-overlay
              (pending-insert (point) "👂")))
      (whisper-run))
    
    (defun sacha-whisper-pending-remove-overlay (text)
      "Add to `whisper-insert-text-at-point'."
      (when (overlayp (pending-ov sacha-whisper-pending-overlay))
        (delete-overlay (pending-ov sacha-whisper-pending-overlay)))
      text)
    
    (defun sacha-gptel-pending-remove-overlay (&rest _)
      "Add to `gptel-pre-response-hook'."
      (when (overlayp (pending-ov sacha-whisper-pending-overlay))
        (delete-overlay (pending-ov sacha-whisper-pending-overlay))))
    
    (defun sacha-gptel-pending-show-indicator (fn &rest args)
      "Show the indicator in the buffer while waiting."
      (apply fn args)
      (pending-insert (point) "💭"))
    
    ;; (with-eval-after-load 'gptel
    ;;  (advice-add 'gptel-send :around 'sacha-gptel-pending-show-indicator)
    ;;  (add-hook 'gptel-pre-response-hook #'sacha-gptel-pending-remove-overlay)
    ;;  )
    ;; This is a test.
    
    
    
    

    sacha-gptel-pending-show-indicator

I used to define voice commands with regular expressions, but I think I've mostly figured out how to use phony.el to define commands using grammar so that I can reuse different components. Let me go play around with this some more.

• on my computer
• on my phone with termux?

OpenVINO/whisper-large-v3-turbo-int4-ov

Current / notes for next time

https://techhub.social/@lispwizard/117401907147288233

https://fosstodon.org/@eludom/117422669292878212

https://fosstodon.org/@eludom/117422669292878212

https://codeberg.org/eludom/home.public.elisp-public/src/branch/master/elisp-public/timer-in-modeline.org

(defvar my-talk-start-time nil
  "When non-nil, the time the talk started (via `current-time').")
(defvar my-talk-timer nil
  "Timer object for updating the mode line.")

(defun my-talk-elapsed ()
  "Return elapsed time string, or nil if no talk is active."
  (when my-talk-start-time
    (let* ((elapsed (float-time (time-subtract (current-time)
                                               my-talk-start-time)))
           (minutes (floor elapsed 60))
           (seconds (floor (- elapsed (* minutes 60)))))
      (format " [talk: %d:%02d]" minutes seconds))))

(defun my-talk-start ()
  "Start the talk timer and enable mode-line display."
  (interactive)
  (setq my-talk-start-time (current-time))
  (unless my-talk-timer
    (setq my-talk-timer
          (run-with-timer 0 1 #'force-mode-line-update)))
  (message "Talk timer started."))

(defun my-talk-stop ()
  "Stop the talk timer and remove the mode-line display."
  (interactive)
  (setq my-talk-start-time nil)
  (when my-talk-timer
    (cancel-timer my-talk-timer)
    (setq my-talk-timer nil))
  (force-mode-line-update)
  (message "Talk timer stopped."))

Probably just a quick session working on Emacs News today.

2026-10-05 Emacs news https://github.com/Daskeladden/canvas-browser

Mode for streaming

I'll work on Emacs News. If there's time, I can show some of the other speech commands I've been playing around with.

• "remind me to…" now saves a screenshot
• "open …" works elsewhere
• "comment …" works in code and subtitle files

Current / notes for next time

https://mastodon.social/@cks/117322231468097911

2026-09-28 Emacs news https://www.youtube.com/watch?v=HqV_l5lry00

https://github.com/ahyatt/llm/blob/decision/llm-typesafe.el https://www.reddit.com/r/emacs/comments/1ws60sw/llm_package_support_for_jev_layastyle_quick/

##yayemacs on libera.

2026-05-25 Emacs news

• Watch on YouTube

<2026-05-25 Mon 13:00>–<2026-05-25 Mon 15:00>

Just hanging out and working on Emacs News

shapr@recurse.social - @sacha Is your org flow written up somewhere?

Like, how you organize your files and how information gets in and out?

and, since I felt a bit cringe about using C-s, how do you find information in your org files?

Emacs Chat 21: Amin Bandali

Mode for streaming

• Is there a way to make headings (ex: Org Babel result block, custom link output)? I'd love it if the headings could get picked up by TOC. Org Babel block reevaluation ends up with duplicates if the result has headings.

Upcoming episodes

• [2026-05-25 Mon 21:25] Sacha and Prot Talk Emacs: May I recommend… sach.ac/yay-emacs
• [2026-05-25 Mon 21:26] Sacha and Prot Talk Emacs: May I recommend… YayEmacs.com
• [2026-06-08 Mon 22:52] Sacha and Prot Talk Emacs YayEmacs.com
• [2026-06-08 Mon 22:52] Sacha and Prot Talk Emacs YayEmacs.com
• [2026-06-08 Mon 22:52] Sacha and Prot Talk Emacs YayEmacs.com
• [2026-06-08 Mon 22:52] Sacha and Prot Talk Emacs: YayEmacs.com
• [2026-06-08 Mon 22:52] Sacha and Prot Talk Emacs: Under-appreciated Built-ins YayEmacs.com
• [2026-09-07 Mon 07:50] Emacs Chat en français
• [2026-09-07 Mon 07:50] Emacs Chat en français
• [2026-11-03 Tue 11:12] Panicked
• [2026-11-03 Tue 11:12] Panicked
• [2026-11-03 Tue 11:13] Panicked
• [2026-11-03 Tue 11:14] Panicked
• [2026-11-03 Tue 11:23] Panicked
• [2026-11-03 Tue 11:23] Panicked
• [2026-11-04 Wed 09:39] Panicked

• Now I can have shortcuts like delete-other-windows-vertically
• Emacs News
• my Emacs configuration
• bug-hunter: Ooh, pierre-rouleau/elisp-bug-hunter has my header/footer code and some new fixes, let me try that
• probably should put together the which-key-display-prefix patch
• sachac/learn-lang: Some tools for learning languages with Emacs - Codeberg.org
• sachac/speech-input: Experimenting with a speech-based interface for Emacs - Codeberg.org

• Boom?
  • Tonor T20 CAD 39.99
  • InnoGear 2026 Low Profile Arm
• Shock mount?
  • Auphonix Shock Mount For Blue Yeti Microphone?

so that it lets me select the link and prompt for the summary right away

Move the functionality of emacs-news-summarize-mastodon-items into it

file:///usr/local/share/emacs/31.0.50/etc/themes/newcomers-presets-theme.el

Also yayemacs.com/news and emacslife.com/news

Current / notes for next time

Emacs Chat

I can't rely on recording true in the getMeetings. I might need to actually talk to the browser through an extension.

(use-package browser-gt :config (browser-gt-start))

(plist-get (browser-gt-request "EVAL_IN_ACTIVE_TAB" '(:code "document.title + '|' + document.querySelector('div[data-test=recordingIndicator] > div[aria-label]')?.getAttribute('aria-label')")) :result)

(plist-get (browser-gt-request "EVAL_IN_ACTIVE_TAB" '(:code "document.title")) :result)

I might not be able to use browser GT to connect to multiple instances.

okay, tentative plan is I'm just going to have to remember to start the recording each time. Which is okay because I need to rearrange the windows anyway. Although the rearranging of the windows happens on the stream. Can I get the stream user to start the recording?

If I use the the mod code on the stream user to start the recording, then the mod code will display on the screen briefly, which is not what I want.

I don't think the server reliably stays logged in. I can't distinguish it based on the window title alone.

I would like to be able to automate things like the recording of the session without leaking the mod code. The server can connect without using the mod code. The streaming will use a regular user. I wonder if I can use Firefox to directly connect to it. Alternatively, I could just have a reminder walk me through the steps. So the key things that need to happen is I need to move the the layout anyway every time I start streaming, and I need to start the recording.

The layout needs to happen in the VNC session. The recording needs to happen in my own session, which I'll have anyway.

okay, I can get the number of participants for the meeting. I need local shell script also to open the BVB room.

Hey, what if I use Firefox to open it? Then can I use spookfox to eval the recording status and possibly flip it? No, it's not activated. Can I use xdotool to find the window?

(shell-command "wmctrl -x -a bbb-listen")

moz-extension://a66f3807-6a57-495e-b4df-c9743e46095e/_generated_background_page.html

file:///home/sacha/bin/bbb creates a window with a specified class

okay, I don't know if I can get the actual recording status unless I use webhooks, which I don't think I can do at this moment, but I will just have to do the recording manually. I have a way now to create a Firefox window with the right class for and position them so that I can have BBB and the etherpad already set up on the workspace and I have a way to jump to the workspace from my overview. I also have the number of participants so I can get a sense of the the activity there. And the next steps would would be to set up the audio devices so that I always have my microphone going in. And the output going to a separate audio device that, that can go out to my headphones with some audio shaping or audio location.

@SLURMInitiative