Hi everybody,

I'm currently mocking up a multi sound event in FMOD where I've edited a few variation of an existing line of (game) commentary to create some more variations. I'm doing this to maximise the use of the existing stock of recorded commentary and make it not so repetitive for the listeners. The idea is to add some pitch variation to these extra short phrases and the extra variations will cycle through randomly.

More specific:

I have made three suitable audio files from one line of dialogue and have imported them. With these settings

Wondering if anyone has done something similar or if this looks like a good way to go about it?

Thanks,

Kim

That's a reasonable approach, but don't fool yourself into thinking that this will make the commentary less repetitive than it actually is. Hearing the same words over and over again is going to feel repetitive even if it's a different recording every time.

@a light breeze

Thanks for your reply and caution. Yes, without recording additional commentary the repetition definitely is still a problem.

I'm quite interested to find out whether something of an adaptive music design might apply to commentary systems and if so, to what degree it is or has been satisfying for people playing the game. I'm in the early days of having a think about the best approach…

• Pitch shifted voice is likely to sound strange because formants are shifted too.
• Wouldn't recording different phrases be more effective and efficient? Much more variation, higher quality (no damage to your recorded audio), comparable effort (more recording but less editing).

Hey thanks,

• Yes that true - simply pitch shifting voice (original dialogue files) may sound too strange to be usable. I'm a few weeks away from being able to hear what I'm attempting with smaller sized files so I'll return to share how it goes.
• Recording new and different phrases is always preferable. I'm exploring these ideas really, as options from smaller projects or where it may be difficult yet preferable to have the same voice artist return to re-do the phrases. Where the vo-artist may have a unique sound but perhaps where the audio budget might not stretch that far.

You should probably prioritize reducing commentary repetitions from the “demand” side, by having your virtual commentators make less comments so that they are repeated rarely and at long intervals.

For example, typical comments for many sports only need simple remarks that can be assembled from very small fragments that are unlikely to sound repetitive because they are trivial, like announcing the score, mentioning which player has the ball (with a few variations and synonyms of the name) or what they are doing (e,g, “X passes to Y") with infrequent smart remarks that are the only audio fragments which could be repetitive.

@LorenzoGatti This is a really helpful and I hadn't thought about reducing comments from the demand side. The shorter infrequent comments that aren't context dependent I'll be able to record. In addition, I can always spend some time with the ambient/environmental sounds so that they are sonically enjoyable for the person playing the game, as there would be more space between the comments. Thanks

This is really an interesting topic. As far as I know, games usually prefer different audio files to using tools in middileware. This is the case for random sounds. The pitch effect has many down sides. It shifts the spectrum and makes it lose high frequencies. Also, for many sound with a notable tone, like the ring sound in a sword clang, pitching it would make it like music melodies, quite weird. To make something less repetitive, 3 random samples are enough in most cases. You need more, of couse, for frequent sounds like footsteps.

For dialogue, one recording is enough for each sentence. They do record 2 alternatives, in some MOBA games.

And I notice in some AAA titles, they use different pre-rendered sounds for the same SFX rather than changing it or blend multiple tracks in middleware. Sometimes it is because of soundbanks management, but the main reason is that, sound design is more flexible in DAW and the approach is less RAM and CPU dependent.

@glenn_zhang Thanks for including your thoughts about this topic. Yes, I agree, the pitch effect on anything metallic would sound, as you've put, quite weird and the voice samples don't respond well to pitching. I have been able to edit the existing dialogue assets to create shorter samples that I'll be able to hear as a series of general/non specific reactions. Fmod has been quite handy to do this as I can hear how they sound as they cycle through in a playlist set to random in an event/multi-instrument. Some samples seem quite abrupt and I'll be able to whittle them out when I hear and test how they sound in the game. At this stage I may not even need to pitch the samples and as as you've mentioned, editing in a DAW particularly for voice, is much easier for careful edits of vocals.