Apologies for my lack of knowledge about the terminology of audio editing; I’m not a pro. But I think what I want to accomplish should be possible through one effect or another.
I have some audio that has a lot of reverb; it was recorded several feet from the interview subject, in a small room. As a result, all of the vowel sounds in the middle of words are very loud and sound jarring.
What I would like is for the vowel sounds, or the sounds in the middle/end of words that include the most reverb, to be decreased in loudness in comparison to the beginning of sounds. Perhaps this could be accomplished through some kind of effect that reduces loudness with a delay, waiting a few milliseconds into a syllable before kicking in? Alternatively, something that increases loudness at the beginning of syllables would do the trick; then I could lower the amplitude of the entire recording and the beginning of each syllable would be comparatively louder.
What I’m looking for essentially is something that makes the sound more crisp and reduces the perception of the reverb by lowering the volume in the middle/end of syllables where the reverb is most apparent. Pleased see the attached image to get a sense of what I’m talking about. All the highest areas in the waveform are in the middle of sounds. I want to mitigate that.
I’ve played around with other techniques to reduce reverb but nothing is quite getting me what I want. Surely the effect I’m trying to achieve is possible? My instinct is that a compressor effect is the way to go, with a setting to adjust the attack time, but I could be getting the terminology all wrong. I’ve played around for this for hours and I’m about at wit’s end, so any help would be incredibly appreciated!!!
The reason nobody has been racing to your rescue is there is no rescue.
Echoes/Reverberation in a recorded performance is permanent. It’s such a common legacy problem, it made it to number one on this list.
In general, room echoes are the performer’s own voice arriving at the microphone more than once. Once from the “real” voice and then multiple thousand times, bouncing from the walls and ceiling. The bounces are late and varying volume.
So the job is to remove the performer’s voice from itself.
The echoes are always of lower volume than the initial performance, so you would think you should be able to sense that. Correct. That’s what the Noise Gate plugin does (attached).
noisegate.ny (5.52 KB)
But Noise Gate doesn’t “know” what a voice is, and will shear off theatrical emphasis and the beginnings and endings of words. It also doesn’t allow the performer to bob and weave in front of the microphone. Give it a shot. You may get lucky. You may get it to “tighten up” the performance enough to be presentable.
I understand there are software packages that claim to deal with echo and reverb, but they’re not free.
You can post a sample of the work on the forum. Go out to 8 or 10 seconds and include some intentional silences. Do not post any work that has been processed or anything you already tried to help.
Also beware of number three on that list. That’s another New User problem. Sound files are not automatically MP3. MP3 is a special compressed sound format that makes convenient, small sound files by rearranging musical tones and leaving some out. So it damages the performance.
If your work is in MP3, then you have two problems, not just one.
What was the performance? Religious leader in front of the congregation? College instructor? Those are two common generators of damaged recordings.
Those can be done, but not with gear the home user is likely to have.
That’s another silly rule. By the time you realize you need it, it’s too late.
I don’t know of a tool that can recognize vowels from consonants and can parse individual words such as you want. There may be some Artificial Intelligence tools that “know” what speech is and how to manage it.
If this is a video, there is a Hollywood Solution. Get the original performer or someone that sounds like them read with a good microphone in a super quiet room and substitute the voices. That’s “Looping” in Hollywood Speak.
Thanks for the feedback Koz! I’m new to audio editing but know a little bit about recording. The issue here was the filmographer. We asked for either a lav mic or a boom to record an interview. What he brought was neither, but we at least got him to place it near the subject. Then it ended up being unusable because there was massive interference with the recording. So we’re stuck with something he recorded several feet away next to his camera. Needless to say we’re livid and are almost inclined to record the whole thing again, except that we hate to ask that of our interview subject.
I’ve attached a sample of the audio (in Spanish). My issue with the recording can be heard in a few places: where he says “mas” at about 2.5 seconds, especially where he says “cuatro” about 5 seconds in. Every vowel essentially, or, more or less, in the middle over every word.
We want something that would be suitable for professional film quality. Do you think it’s possible to get there from here? Otherwise we’ll give up on this and re-record.
Trebor: much thanks for the tip! I’m going to try expansion
Thanks for the sample. Spanish doesn’t matter to the problem.
This has already been processed, right? It doesn’t have any background noise. The sound at about 8.3 seconds is digital dead quiet. Or was that a setting on the recorder?
You probably have the worst production problem. It’s aaaaalmost good enough.
I have no opinion. It’s up to you and the filters.
Lol tell me about it! I could spend all day playing around with this audio, and yesterday I did. This didn’t do anything to make me less angry over the whole situation.
What I sent is what my filmographer sent me; I amplified it a bit. He’d been trying to fix it. Perhaps I should get the original audio from him.
Probably won’t help. It’s just that when we’re in repair mode, it rings bells when there is any hint of processing or that somebody got there first.
I have a portable recorder with normal settings and then a page of special effects settings. High Pass Filter, Auto Volume Control, Zoom Mic, VCVA, etc, etc. I have most of it off.
It’s a wide stereo recording. If you delete one of the stereo tracks, to make it mono, it will sound less roomy.
Then remove as much of the bass as is acceptable, (that’s where most of the room-everb is).
(You were right: that particular sample needs compression).
Thanks for the tip! Playing around with that now. The Dynamic Mirror plug-in ended up being exactly what I was looking for so props for that. The audio at least doesn’t sound so glaringly bad now, if not great.
Our consensus is that we’re going to keep this audio for now but try to reshoot the interview later on when time and budget permits. We feel that it’s worth it for something we want to be cinema quality
The main problem I see in this recording is the unnatural and very strong emphasis / accentuation of some of some syllables: “mas”, “cuatro”, “colectios” (… or whatever words the speaker said and I am not able to hear properly ).
Another way of improving the quality of this recording would be a manual lowering of the volume of these parts. Tedious, though. Depends also on the length of the recording…
).