I completed a 60 minute interview with a gent who has an accent, a minor stutter and major voice volume issues. (The volume stays normal for a while, then goes very very soft and then very loud) Listening is a chore. The accent is not a showstopper. However the voice tone is garbling words and the stutter is distracting. I am trying to edit the voice to an even tone. And am struggling to find the right combination to make this a less painless exercise. His message is very important, otherwise I would not spend so much editing time on this. Does anyone have experience successfully editing stutters and voice cadence? Thanks so much
Here is a small sample of the recording I am playing with
Did you fix your recorder since that performance? This sounds like a recorder that’s trying desperately to “help you” with your voice performance. It can’t figure out which parts are real voice and which parts aren’t. When it gives up are where you get the holes in the sentences.
I don’t know any way to repair this in post production other than the Hollywood Fix.
Write down the words and have an actor read them in a studio. If you do it right, the interviewee will have no idea what you did.
I am not sure I understand your direction. What does fix your recorder mean? I use riverside fm as my recording platform. This is not the problem of the platform. I’ve used that platform flawslessly for three plus years. What you are listening to is the man’s actual speech in raw time.
This was a live interview with an expert I do not personally. The gentleman has at least two speech impediments. A stutter and a voice volume that is not maintained consistenty. Plus he had a slight accent – not a showstopper, but enough to get in the way periodically.
I appreciate your suggestion to get an actor. However, there is little in the budget for that kind of thing.
I have repaired the sound somewhat. By inching through each second and literally amplifying his low tones and removing his stutter. As you can imagine, it is extremely slow. This was a 60 minute interview am I’m not sure I have the bandwidth for what will no doubt turn into a more than three day exercise doing what I am doing.
Your posted sample appears to be very highly processed, honky, talking into a wine glass distorted. I don’t think the human that spoke those words actually sounds like that.
That’s the recorder—or recording process—that badly needs to be fixed before you do any more production.
That’s what I’d be doing. I don’t know of any push-a-button process that can fix that kind of damage.
Since you have tons of editing ahead, you need to know that Audacity creates a backup copy of the whole show each edit you make. That’s its Edit > UNDO process. If you make a mistake, it doesn’t actually undo the last edit, it just plays back the last show. Your machine can fill up with all those copies and crash.
Periodically Edit > Save a Backup Project under a unique name.
This is where one of the AI people posts with a GPT filter for this.
It’s possible to improve intelligibility somewhat with equalization and compression …
(before-after)
But his speech impediments combined with noise/echo-reduction “audio enhancement”, (at his end), is a worse-case-scenario.
See? I’m not the only one that thinks the voice work is highly processed.
I bet that’s not him. I bet that’s the processing.
I get the unmistakable feeling of walking into the middle of the movie. This clip is so badly damaged that I would want to execute a totally plain, ordinary voice recording in whatever room they’re using and analyze that. This clip has way too much “help.”
It’s not rocket surgery. If they have an iPhone, make sure Voice Memo is set for Lossless in Preferences. Lay the phone on the desk with its butt pointed to the speaker, press record, and read something.
OK, so word-by-word editing it is. There’s still no automatic way to correct that I know of. It still sounds milk-jug/wine-glass processed and I bet all this would go smoother if it wasn’t.
Android phones seem to have Voice Recorder which is a similar product to Apple’s Voice Memo. It operates lossless and can save work as perfect quality WAV. It is suggested it has noise reduction. We’ll see about turning it off.
It’s free-ish. It has ads. Does this mean it actually burns ads into the show? I wonder how they manage that. Every so many minutes? Once a recording? What if you’re in airplane mode?
Thank you @Trebor. That is EXACTLY what he has. And thank you for sharing this. You’ve given me the exact name and details. Frankly, I was amazed he had the confidence to do the interview despite his issues. I want to give him the platform for his knowledge and content
For future reference, he should turn off all the typical “audio enhancements”, be they on Windows or Zoom, as they expand the dynamic range, when his voice already has an unusually wide range of volume.
Dynamic-range-compression would reduce the volume variation in his voice, that compression can be done in real-time (live) with hardware or software.
Thank you @Trebor. I am very grateful for your help.
After playing around and experimenting with Audicity, I adjusted the compression settings to threshold being 40 down and make up gain being 25 up. The voice is now reasonably legible. And you’ve saved me hours and hours of work. Thank you.
