I’m trying to add more bass to an audio file that has way too much treble. It’s an old recording of someone talking made on a casette in the '80s.
My problem is that Equalization doesn’t help so much. The “voice” datas seem to be only in the “treble area” on the spectrum graph.
I mean: when I amplify the low frequencies, there’s no change to the voice. I notice a change only when I amplify the mid-high and high frequencies.
Is it possible to “move” the spectrum to the left, so the datas located in the high frequencies would be moved to the low frequencies…
Could you post a screenshot of the spectrum (Analyze > Plot Spectrum), or alternatively, add a short sample (just a couple of seconds) as a WAV file to your reply. Files (less than 1 MB) can be attached to your post (see “Upload Attachment” below the compose message box).
Realistically, there’s only so much you can do. Equalization is probably going to give you the best, most natural, results.
You can sometimes use filters & effects to turn a good recording into a great recording, but you usually can’t turn a bad recording into a good recording… And sometimes ‘the cure is worse than the disease’.
Is it possible to “move” the spectrum to the left, so the datas located in the high frequencies would be moved to the low frequencies…
Yes… You can try the Change Pitch effect. Maybe try -25% to -50% change. And you can try mixing the pitch-shifted copy with the original, and you can try using different EQ or filtering the pitch shifted and original copies before mixing. But, whenever I’ve tried this kind of thing it has just not sounded natural…
Below 200 Hz there is only noise. The person speaking does not have a deep voice so there are no bass frequencies (other than noise) to bring out.
I’d disagree about there being too much treble. There is too much low frequency noise, and frequencies in the range of about 400 to 500 Hz are overemphasized.
I would recommend using Equalizer settings something like this:
That sounds like noise reduction has been applied, which may be responsible for the unsatisfactory equalization.
Better quality may be possible from the raw original, (i.e. before any noise reduction has been applied).
steve:
There is too much low frequency noise, and frequencies in the range of about 400 to 500 Hz are overemphasized.
Can you please explain how you came to that conclusion? Did you just noticed it by watching the spectrum, or by listening the sample? Maybe I could find documentation about how to use equalization…
Yes it is a child, but the voice isn’t authentic. It’s sounds like a different child. Some other recordings are more authentic ; they have been made on a different radiocassette player.
(Yes, noise removal has been applied to eliminate the background sound of the cassette recording).
I could tell that there was no meaningful content below 200 Hz by removing everything above 200 Hz using Equalization like this:
I could tell that the 400 to 500 Hz range was over emphasized by the sound, but it is confirmed by that large peak on the spectrum plot.
No, not necessarily. Different sounds will have different shapes. Some sounds will appear quite bumpy while other sounds will have a more smooth shape. However a voice recording will not usually have a big bump like there is in the sample posted previously.