Efficient ways to handle Sounds at Runtime?

[dragonshield12345 (Kronaby)]

I’ve been looking into how to better handle sounds within my game, which is going to be multiplayer and quite heavy on sounds. There doesn’t appear to be any concise answers as to the best way to handle sounds during runtime, but I have found some useful forum posts that specify more efficient methods.

• Serialising the sound data and creating the sound instances during runtime as needed.

• Storing the sounds where they’re needed and using PreloadAsync() so there’s no delay on playing the audio.

• Using a ModuleScript in ReplicatedStorage to handle sound playback on both client and server.

Whilst these are all efficient, I’m unsure what the best way to handle everything in one would be. It seems less efficient to create new instances for small sounds played often, and preloading a bunch of small sounds often also doesn’t seem the most efficient use of server memory.

My current idea would be:

• Keep small client sounds such as UI/SFX that are played often in the location they’ll be played from so they’re always loaded.
• Any larger audio like music tracks is serialised and stored in a ModuleScript in ReplicatedStorage so both the client and server can access it, and then preloaded prior to being played.

Is this the right way to approach this topic, or is there a more efficient way to be handling audio during runtime?

Extra: I also plan to have creatures and other assets that are replicated often. What’s the best way to handle each instance also needing its own audio?

[dragonshield12345 (Kronaby)]

In case it helps anyone else, here are the links to the other forum posts that explain their methods in greater detail:

How to store sounds and music in your game correctly

By far the most efficient and scalable solution is to serialise sound data (value instances or a folder with attributes are suitable) and create sounds during runtime when required. If they’re played consistently in the same location in the DataModel then it acts as a cache of sorts. Clearing the sounds out of this location will remove them from the cache and achieve the same effect of this post. This method beats the one mentioned in this post by far: You can make a ModuleScript to handle s…

Designing for memory and performance: instancing sounds or keeping them in the DataModel

I don’t know the inner workings, so take my advice with a grain of salt. I might be wrong on some details here. Assets that you upload to Roblox, such as decals, meshes and audio are all stored on the website. When you insert any of these assets into your game, you have to download the referenced asset from the website using its ID. So when you instantiate a sound which isn’t yet present in memory, you need to download its data from the website, which can take a brief moment and take up network…

[Zomebody]

I’d say your approach is very solid. There are a few key things that are important to remember:

1. Assuming Roblox works the way I think it does, the server doesn’t actually need to load audio. There’s no person on the server that can hear audio. All the server sees is the string that contains a reference to the audio that is stored in the Roblox cloud. Only clients will actually (down)load the audio data, so it’s fair to assume that audio will have a negligible impact on server memory.

2. For clients it’s most important to stay within the “budget” of memory that a client has. Mobile phones are usually the most limiting factor as the cheapest modern phones have roughly 1-2GB of memory. According to the files on my own PC, music files can be multiple MBs in size, while audio effects are usually less than 100KB. With this in mind you can safely load and play dozens of songs and still have enough space left to load your game’s images, interfaces, code, and so on. But if your game is heavy on music, it would be smart to take precautions.

3. Although I’m not 100% sure, I believe that sound instances themselves take up almost no memory. Only once you play a sound instance will Roblox fetch the data from the cloud, store it in your memory and play it (which is why the first time you play music there’s sometimes a delay). So whether you have a folder with a 100 sound instances or a ModuleScript with 100 id strings is up to personal preference.

There are still a couple unknowns that I haven’t found any documentation for. For example what happens when you start running out of memory? Does Roblox automatically wipe audio data from memory that hasn’t been played in a while?

That said, I think your approach strikes a good balance between convenience and performance. Keep cheap sound effects loaded for quick response times and flexibility, and only load music when it’s needed.