Youtube audio distortion

[asdv]

Just started using Youtube to promote DVD sales but I can't get good audio quality - it's the narration that sounds a bit rough and distorted after upload. Most people with poor sound sytems on the computer wouldn't notice, but it isn't as good as it should be given the current standard of H264 and AAC technology.

It is a known issue - there are posts on the Youtube forum about it but none of the suggested fixes works for me - mainly lowering levels, particularly checking and lowering high energy low frequencies (most NLEs don't meter audio accurately).

I'm uploading using their recommended guidelines - AAC codec, but even uncompressed audio produces the same result after upload.

Any one else dealt with this ?

Forgot to add the link to the problem audio clips:
http://www.youtube.com/user/WildlifeInCloseUp

[Gavin Gration]

Are you talking about the "ess" type noise on the narration?

[asdv]

No, it's more a general across the range distortion I am hearing.

Sibilence has been an issue with this narration but was supposed to have been fixed to some extent by a post house (longer story I won't go into right now).

[rogs]

I don't think any 'distortion' is necessarily the fault of the you tube encoding
I think you'll probably be better off by reducing the dynamic range of the narration.
I take it you are using your own voice, and unfortunately you have one of those (very pleasant) voices that doesn't record well for narration purposes!
Not trying to be rude here, (honestly!) but there are some aspects to your voice that make it a difficult one to get 'professional' sounding 'voice over' type results.
The fundamental resonance has few harmonics, and so intelligibilty will rely on the so called 'non-voiced' harmonics -essentially the sibilance.
In your case, that sibilance is not well dealt with by the standard de-essing technique. It has very dominant content between 3.2 and 4KHz, and those 'peaks' will tend to define the maximum amplitude of the signal, making the overall loudness level much lower.

I've taken the first 30seconds of your audio from the you tube clip and tried to make it 'louder' without actually increasing the maximum output level at all. In fact it's about 3dB lower!

I've posted a small mp3 file of the excerpt here to show you what I mean.

Unfortunately, I'm no 'voiceover' expert, so I'll leave it to others to describe more professional ways of improving your audio, but I do feel that a professional reduction of the dynamic range, coupled with some 'prescence' enhancement could make quite difference.

It's a very sophisicated business, audio 'sweetening' --just look at the price the broadcasters pay for Orban or Omnia equipment, to do just that!!

[asdv]

rogs, thank you for taking so much trouble and apologies for slow reply.


Speaking has never been something I relish - this film did require it unfortunately and getting a half decent performance out of me was like getting blood out of a stone.
I've been aware the narration was not what it ought to have been technical-quality-wise but unfortunately I did not know that at the time it was recorded so I have little re-course to the studio that did it for me. They also lost the v/o only track recording so I have only the final mix with all effects included ! The file you made certainly sounds better and demonstrates the dynamic range point well. I rather doubt the studio did any compression at all. I actually came to the final online session with my own narration which was condemmed as having too much sibilance.
I think I shall have to find the cash and will to record it yet again - I am already facing a court action for a bad quality DVD authoring job I refused to pay for - every element of the production I took outside for help seems to have got me into trouble and wasted expense.

I am still hearing similar type of poor voice quality on some other youtube videos (on some but not all other people's channels) so that problem is still a puzzle to me.

[rogs]

I'm surprised the audio post house presented you with a file with such a wide dynamic range, especially if it was always destined to go 'online' with YouTube?
One might think they would realise the majority of those likely to access the video are going to be listening using typical 'computer' type speakers, and a wide dynamic range spoken narration is always going to have 'loudness' issues in that audio environment, IMHO!

It's shame you don't have the the raw vo file, because trying to alter the dynamic range of the final mixed file is always going to have limitations. In this instance, the background ambient noise signal is obviously important to the whole thing, and trying to apply dynamic compression to that kind of file is almost certainly going to make the background signal 'pump'. Not good!!

I've had a go at remixing the whole file, with the specific intention of reducing the resonant sibliance, increasing the perceived loudness of the whole thing without increasing the actual maximum level, and adding a bit of 'presence' to improve the intelligibility of the narration.
As I said in my previous post, I'm no professsional voiceover expert - my audio skills (such as they are!) are much more concerned with the reproduction of spoken messages in emergency situations - but I do like a challenge!

I've posted the complete modified file here if you're interested.
It's in a similar PCM .wav format to the soundtrack I extracted from your downloaded file -and it is exactly the same length, so if you want to use it, you're more than welcome.
(Because it is a .wav file, and not recompressed, it's about 28MB.)

As I say, there's only so much you can do with a 'mixed' file, but I am surprised they didn't try and process the narration before the final mixdown!

[asdv]

That does sound better - thank you very much.
I will try uploading that version instead and see if it goes through the Youtube processing better.

[rogs]

I shall look forward to seeing how the 'you tube' encoding affects the modified audio file.
If it turns out OK, then I'm glad I have been able to help a bit.

I find it very annoying that some 'professional' audio houses seem to think that everyone has perfect hearing, and that we should all revel in the wonders of high definition digital audio, and marvel at the huge dynamic range available from modern recording techniques.

In reality, there are many people - especially older folk - who do not have perfect hearing, and find it much easier to listen to recorded dialogue, when the dynamic range is restricted.
Listen to the intelligibility of some of the dialogue soundtracks from some of the 1970's and 80's TV shows. Programs like 'Columbo' and 'Quincy ME' for example.
Not very 'pretty' audio by today's standards perhaps, but it's certainly easy to understand every word, without forever using the remote to change the volume every five minutes!

Of course, they were made like that because the program makers knew they were going to be played through rubbish TV speakers. But many of today's computer speakers aren't much better!

Same thing applied to the cinema soundtracks for 'Pathe News,' and the like.
Acoustic responses based on the notoriously inadequate ' Academy Curve,' coupled with noisy optical sountracks meant that serious compression was required for intelligible dialogue. Even deliberate clipping was used in some cases!
But you could understand every word easily!

Maybe some of today's trendy audio dialogue recordists might take a look at some of the work of previous generations.

Especailly with regard to wide dynamic range --just because you've got it, doesn't mean you have to use it!

[Rob James]

Dialogue is king! If you cannot hear it you've lost half the audience.
A big part of the problem is 'trendy' acting techniques (mumbling) which have been the bane of my life for years.

Once recorded Billie Whitelaw doing a VO for a Beckett piece. She was so quiet the studio base noise was louder - nightmare!

The director wanted it that way and he who pays the piper...