Skip to main content
Participant
April 24, 2024
Answered

Premiere Pro Enhance Audio is adding audio artefacts to my clips

  • April 24, 2024
  • 14 replies
  • 6043 views

Dear Community - please help!

 

When I use Enhance Audio on my clips it adds audible 'gremlin' artefacts to the clip. They are not always visible on the waveform but here is an example where it is visible.

 

The first clip uses no enhancement, then I set it to 2, 4 and 7.

 

 

In this particular case, it would appear to be trying too hard to amplify/improve barely audible background noise or mic rustle as it gets worse the more I 'enhance' the audio. But this is the very opposite of what the effect should be doing, surely it should be squashing wind, rustles, noise?

 

This has also happened over spoken/ambient audio, and not affected the waveform - which is scary, as they are not always picked up in review!

 

This is happening over multiple Macs and always on the latest version of PP. The error is repeatable using completely fresh projects, sequences and clips. I have had to stop using Enhance Audio for fear of missing some of these glitches and them going live.

 

Here is the clip info:

 

Type: MPEG Movie

File Size: 6.49 GB

Image Size: 4096 x 2160

Frame Rate: 25.00

Source Audio Format: 48000 Hz - Compressed - Stereo

Project Audio Format: 48000 Hz - 32 bit floating point - Stereo

Total Duration: 00:05:47:02

Pixel Aspect Ratio: 1.0

Alpha: None

Color Space: Rec. 709

Color Space Override: Off

Input LUT: None

Video Codec Type: HEVC 10 bit 4:2:2

 

Audio:

Audio track 1 contains 1 type(s) of audio data: Audio data block #1: Format = 16 bit - Stereo Rate = 48000 Hz Compressor = sowt

Correct answer Graham Hutchins

Yep, add a fairly loose dynamics/autogate effect to the audio clip > render and replace > enable enhace speech seems to be working.

14 replies

Inspiring
April 29, 2024

Yes, I am having this same problem. In spaces of total silence in the audio (a Zoom call recording), Enhance Audio is adding (hallucinating) little samples of speech from a different part of the clip. Once it has started hallucinating a certain sound in a silent place, that same sound sample is repeated again in other silent parts of the clip.

 

I hope Adobe realizes that Enhance Speech is effectively broken right now. If it's going to add random sounds to my audio how can I trust it?

 

I have hours of online learning material that could benefit from Enhance Speech but I can't afford the additional time needed to fully check all exports for random sounds the AI improperly added.

Remote Index
April 24, 2024

"magically remove noise and improve the quality of your dialogue clips so they sound like they were recorded in a professional studio."

 

Unfortunately, what you've quoted is from the documentation / user-guide which one would hope would be as free of marketting speak as possible.

 

Note to Adobe on documentation: if the user guide is telling me something is done "magically", you're probably not doing a great job of explaining what your software does or how to use it.

 

R.

Participant
April 24, 2024

Thanks for responding R.

 

Yeah, it's a shame that it does a good job with speech but doesn't handle silence well! But as the marketing says "magically remove noise and improve the quality of your dialogue clips so they sound like they were recorded in a professional studio." 

 

If you listen to my clip again - the silence at 7.0 enhancement is great - it's just chucked in new sounds! And I like your tip about dipping the audio when there is no speech, but this could be a very manual fix for a supposedly time-saving filter.

 

And, it's still dropping these glitches into my standard dialogue, hence me not being able to use it at all.

Remote Index
April 24, 2024

Hello Mr. Nathan EB

 

"In this particular case, it would appear to be trying too hard to amplify/improve barely audible background noise or mic rustle as it gets worse the more I 'enhance' the audio."

 

While I have little experience with it, you seem to be referring to the Enhance Speech tool.  I think you have correctly described what is happening.

 

The Enhance tool in Enhance Speech is designed to 'clean up' and improve voice recordings. As such, it will be very sensitive to voice sounds (or sounds that are voice-like), and attempt to 'improve' them.

 

I'd suggest removing the passages with background noise/voices (and adding room tone to fill the gap) in your example above, leaving only those passages with dialog you wish to use, supported by room tone for continuity.

 

But as always, the first recommendation will be to focus on getting good clean sound in production.

 

R.