QUESTION · MEDIA PRODUCTION Q&A

Does the panel run Whisper locally for transcripts, or is it worth paying Descript or Riverside for the cleanup and speaker labels?

ASKED BY · Paul Valhusenos · TEXAS · 18 SEP 2026 · ▶ 00:42:36

THE PANEL'S ANSWER

Mac Whisper works well locally with speaker detection; DaVinci Resolve Studio's built-in transcription with speaker detection is faster and more searchable for video work.

David BradyPANELIST

Whisper runs locally well, especially on Mac minis; set up one as a dedicated transcription machine and send everything to it.

Fenwick

Riverside can transcribe finished videos, not just recorded sessions, which is a useful feature for post-production work.

Summarised by AI from what the panel said. Watch the moment for the full answer.

IN THEIR OWN WORDS

From the transcript.

Machine-transcribed, so it may contain errors. Every timestamp opens the episode at that second.

42:25They opened it up and they found that there was liquid damage from beer. It had gummed up all the components. And they're like, yeah, we're not going to cover that under warranty. So beware of used DJ gear. Next question. All right. Paul Valhusanas in Texas asks, does the panel run Whisper locally for transcripts or is it worth paying Descript for or Riverside for the cleanup and speaker labels?

42:49Alexander? Well, I've had some experience with Mac. So I use Mac Whisper and I have used it, although I haven't used it in probably about a year or so, but I own a license for it. It works really well. It runs locally. And it does a really good job of separating the different, identifying the speakers and the audio file. However, I find that I'm now,

43:09because I'm mainly working with video production stuff now, I'm using that feature built into DaVinci Resolve Studio. So you do have to pay for Resolve Studio to get this. But it does have speaker detection and speech-to-text translation. And it works exceedingly well. It saves me a lot of time. and I can actually after you know it processes the hour and it might take

43:33like for say 60 to 90 minutes on my m3 max MacBook Pro it might take I don't know five or six minutes to process all the audio but it does a really good job and then it's all searchable of course so it works really well for me to you know if my client says hey did did so-and-so say that one thing at some

43:53point in the episode and I can just type in a few keywords and boom it It instantly jumps to the point in the timeline where that person has said that. And then I can easily grab the in and out points and grab, if it's a good segment, I can grab that as a clip for social media. So that works really well. And that's what I'm primarily using.

44:10David? Yeah, I'm a big Whisper user. I do a lot of translations via it, translations into subtitles, et cetera. But I find running it local is fine if you have, especially with this community, with all the Mac minis that everybody has just laying around. Just set up one as a Whisper machine, send everything to it, and call it a day. Fenwick. You guys are wicked smart.

44:37So just yesterday I had to do a quick transcript, and I don't use the Resolve, and I don't even know what this Whisper thing is. But I figured out, I found out that you can upload a finished video to Riverside and Riverside will do a transcript of something more than just the stuff that it records for you. I thought that was pretty cool.

44:59Yeah, I also dig the, I dig the one that's built into Final Cut as well for the transcripts just because being able to search through your timeline with text was, is, is just pretty darn powerful. I know that all the editors are doing it and it's table stakes now. But if you haven't tried it because you're like, oh, I don't need that, you should really try it.

45:20I was getting ready when we were doing one of the anniversary episodes for Office Hours. I had pulled the original, like the first episode of Office Hours to try to get a piece of it. And I knew something was mentioned and a story was told. And then I was like, found it. Amazing. Hey, I got a question for you. Did we make you smile today?

45:43Did we make you laugh? Did we save you some money? If you're getting some value out of Office Hours, I encourage you to put some of that value back at officehours.global.com. We are a nonprofit. We are largely member supported and we really appreciate everybody who helps us keep this going every day. The folks that work on the show are 100% volunteer, but we do have bills to get this out to you.

46:04So appreciate it at officehours.global.com. Sorry, I lost your audio there. Oh, next question. No, you're good. Okay. Tommy Shanson, St. Paul, Minnesota is asking, I'm looking into using Riverside. Are there any things to look out for that might make things more difficult than need be? I know Chris Fenwick's a big Riverside champion. I am not. But I'm going to throw it to him anyway, even though he didn't have his hand up.

READ THE FULL TRANSCRIPT →
ALSO IN THIS EPISODE

12 more questions
from NDI Mixing (HX-HX2-HX3) and SDI Reliability.

The whole episode →
▶ 00:00:21I want to take portraits and upload them to a watch folder so ProPresenter will display them on the television monitor. Landscape format images work fine. How do I make portrait formatted images show upright automatically?▶ 00:05:53What are your opinions on the EBU's O-Graph for broadcast graphics, and what's the current state of its usage in the industry?▶ 00:09:37Anyone see the Blackmagic announcement for Generation 6 Color Science? It's in the 10.3 camera update. What impact will that have on production?▶ 00:11:24Supertone Clear and Xcentize DX Revive claim to rebuild speech, not just denoise. Is regenerated dialogue ethical for a documentary interview?▶ 00:20:47I have two Sony PTZ cameras. One is an NDI-HX2, the other is an NDI-HX3. Is it okay to use NDI output from both cameras together as inputs to vMix, or should I capture them with SDI?▶ 00:22:21Does anyone know who the cinematographer was for the Apple keynote last week? Shot on iPhone 18 Pro, but this year it looked noticeably better. Much better color and contrast lighting. Thoughts on the production?▶ 00:26:51Google Meet's Gemini note-taking defaults to on for business plans from September 21st for meetings of three or more. Should organizations announce AI note-taking the way they announce recording?▶ 00:34:29Are there categories or specific items of AV gear that you consider okay, at least less risky to buy used, either to save a little or because they offer something not found in the newer versions?▶ 00:46:37I'm looking into using Riverside. Are there any things to look out for that might make things more difficult than need be?▶ 00:49:02I have a project that will require some not complicated graphics and light animation. I'll hire a graphics artist for a day, but would like to explore using AI. Any suggested platforms?▶ 00:52:45During the pre-show, the panel were discussing a video series and show about an abandoned construction site. Series name, please. And its spinoff.▶ 00:55:30Is there a better AI for mechanical or architecture design? Gemini and ChatGPT were useless to me.