Got Music Department?

[Prolix]

certainly cdaudio quality is not needed. I just made a recording using a simple desktop sound recorder, I used the voice lossless preset to save as wav. I then converted it to mp3 using lame with a standard fast preset (probably could have made it smaller.) I can't hear the difference, can you?
wav 604kb or mp3 60kb

By the way  I was using a cheap condenser mike that came with dictation tape recorder. So by all means have somebody with the high end audio equipment post process stuff if you want but pretty standard equipment is good enough for something that is going to run on pretty generic computers. Most people do not have the latest and greatest.

[zanzibar]

The latest and greatest isn't necessary.  The SM58 was invented in 1966, and it isn't a very expensive microphone as far as good microphones go.

I'm looking at the mp3 you uploaded.  It's 160 kbps.  When someone is talking about a poor quality audio file, they aren't typically talking about a high quality mp3.  Yes, mp3s are lossy vs wave, but the method I described will result in much smaller files.  The method is more complicated than simply changing file types.

Something was wrong with the .wav file.  The file was damaged in a very strange way.  Try reuploading it.

[khoridor]

Two questions come to my mind:
- How are you planning future dialog updates? Surely, actors will have to be dedicated for longer than short term.
- (off topic) You said quests go menu-driven. Does that mean list-clicking answers?

Otherwise, I've got only one word for sound immersion: 

[Vengeance]

We are going to use the speex format to compress the files.  MP3 (or Ogg) is a generalized format for music and such, and thus missing opportunities for compression of voice that speex finds.  Speex is a very dominant format, being used in lots of things from Ventrilo to Asterisk.  We are seeing compression ratios of 40:1 and even one of 75:1 with "acceptable" quality.  This is crucial because the audio will be pushed in real-time on demand to each client from the server.

You guys are welcome to obsess about model numbers of microphones but I think it shows that you are missing the point.  As Xillix pointed out, we are not even attempting to get to "pro" quality audio here.  This isn't Oblivion, I believe he said.

- Amateur voice actors
- Uncontrolled background environment (machine fans, etc.)
- Low quality mics
- Compressing the bejesus out of the result

We are doing this because we think it will make the game more fun.  We're offering to have players help us record because we think it will be fun for people to contribute.  If you think it is fun to help write the guide about how to make the best audio, that's fine too, but I don't want to scare off people from trying to record.

If you don't like the end results, you can certainly disable the audio voicing entirely and never have to hear an amateur voice or compression artifacts.  :-)

[miadon]

hmmm then we can play another game, guess who's voice it is \o/

[Prolix]

That wav file used the codec araw according to vlc which did not actually produce any sound for it, it played fine in audacious and xmms, perhaps it was a codec issue. The tool I used to make a recording I also used to make a speex based file which was quite small but the tool gave it a spx extension which I could not find a palyer that would play it except the tool that I actually used. Anyway this was just a quick and dirty example I made for posting.

I could have compressed the mp3 further and the wave file was at 2200khz, 32bit and 705kb/s as far as that goes.

Speex is commonly used in voip programs like teamspeak and i believe skype. It is probably a good choice and obviates the need for fancy equipment.

As far as synthetic voices go, ie festival, I think it might be useful, for kran or derghir voices. It doesn't really matter what kind of license the application has because you can always encode the actual sound produced with a different codec, the one concern might be distinctive voices that may be held to be proprietary. You can however use this tool to create your own voices. You would probably want to for non-human inflections anyway.

I think this is a fine initiative, however it works out.

[Vengeance]

There is zero chance we will be using a voice synthesizer.  Boring robot voices will not achieve our goals.

Speex is part of the Ogg framework, so if you download Ogg codecs for Windows Media Player, WMP will play the spx files.  CS obviously has a new speex plugin now, developed by us, so you will not be required to download Ogg or anything else other than PS to play PS.

[zanzibar]

I could have compressed the mp3 further

A lot further. The file was much too large to be useful I think.

[Mythryndel]

* Mythryndel examines the mutilated corpse of what might possibly have once been a horse...

[Raekh]

I vote for Prolix to contribute/apply - his voice is sonorous and fitting, in my eyes

[elachlan]

Why would we not go for this (http://www.cstr.ed.ac.uk/projects/festival/morevoices.html) it is perfect for what you want (on demand pushed to the client) and you record your own voice samples to build up a profile. No idea why you would think this is like microsoft sam. Sounded like a normal person.

PS. if you do build a profile prolix should be in for sure (has my vote).

[Vengeance]

Umm... because robots can't act?  :-)

[weeman182]

We know that studio mics like the SM58 are the gold standard, but where do other microphones fall? 

Gold standard? Im a recording engineer and by no means is a SM58 good enough by my standards for any professional game. First reason is the microphone was designed for a live sound enviroment and not a studio one. However the Sm57 has been used on micheal jacksons earlyer albums to great effect but this was mostly due to high quality expensive outboard. I know this is counter productive but no one has mensioned condensor microphones yet so. I also would like to help counter this speech problem your having. typicaly speaking the human voice consists of a high amount of mid range frequencies, because of this pc microhpones are only designerd for this range in mind "partly unnatural but possibly effective:". Due to the problem your having i wouldnt consider high end microphones a lickly resolution as their made to capture the full vocal range which you hear daily on single tracks and for the most part people wont own them. Also its highly likely that no one studio will be availible free with enough voice actors. What i propose is geting people to submit a audition via mp3 format at 128kps mono of a pre wrote dialogue from this you can esstablish the quality of sed persons equipment, their voice tonality and the type of enviroment their using for recording which imo will be a big factor in producing a quality product. After which i would assign each actor a spesific role to suite their vocal type. Lower voices may suite larger characters where lighter more well spoken voices may suite more intelliget/smaller characters such as elf's "think lord of the rings". submissions with poor results should just be discarded because they will probably bring down the quality of the product. Im sorry if that seems abit harsh to seem but i think its a realistic. im not trying to tread on peoples feet just offer an opinion. Again its late when i write this so my english may be poor as im vary tired. MAny Thanks John hope ive helped a little xxx

[khoridor]

Ok, so let's say I have a beautiful, deep manly voice with a tiny bit of an accent, and I'm ready to audition.
What kind of hardware do I need to obtain the required quality level? Can someone post a clear, definite answer to that? Anything that plugs directly into a PC, or not?

[Eddyster]

Hi guys (i'm a new prospect 

I used to sell these, and would rate them for their price. This depends on what the dev budget is of course.

http://www.provantage.com/samson-technologies-sac01ucw~7SASN001.htm

Straight into audacity!

Also, whilst I realise that this is no doubt a decision for the developers, I would be happy to produce the sounds recorded using a mic like this (including noise removal, vocal compress and audio compression, etc).

Here is a very short sample of a voice over I recorded and produced for open source game OpenOutcast. This sample is 12kb in size of the OGG format and sounds great in game.

Edd