Sources

Objects that produce a stream on an Output pin.

AudioPlayer

Plays an audio asset. This replaces Sound.

Property	Type	Notes
Asset	ContentId	"rbxassetid://123". Loads at once if AutoLoad is on. (AssetId is the deprecated old name, still shown as "AssetID" in some tutorials.)
AutoLoad	bool	Load when Asset is set. If off, the asset loads on the first Play().
AutoPlay	bool	Start when it first enters the DataModel. Only for players created or loaded locally, not ones that arrive by replication. Mostly for Studio.
Volume	float	0–10
PlaybackSpeed	double	0–20. Changes pitch too.
Looping	bool
PlaybackRegion	NumberRange	Start and stop seconds within the asset. Min = max means the whole asset.
LoopRegion	NumberRange	Loop points inside PlaybackRegion. Min = max means use PlaybackRegion.
TimePosition	double	Playhead in seconds. Read or write.
TimeLength	double	Read-only. Length of the loaded asset.
IsPlaying	bool	Read-only to scripts, but it replicates. Change it with Play() / Stop().
IsReady	bool	Read-only. Loaded and buffered. It can go false under extreme memory pressure.

Method / event	Notes
Play(atTime?) → actionId?	Plays from the current TimePosition. Replicates from server to client. With atTime, the play is scheduled against SoundService:GetMixerTime() for sample-accurate timing.
Stop(atTime?) → actionId?	Stops where the playhead is. It works like a pause: set TimePosition = 0 to rewind. Can be scheduled the same way.
Cancel(actionId) → bool	Cancels a scheduled Play or Stop that has not happened yet.
GetWaveformAsync(timeRange, samples) → {number}	Volume samples over a time range without playing. Use it for waveform UIs or to pre-plan lip sync. Yields.
Ended	Fires when playback reaches the end. Not for loops, and not when you call Stop(). To catch every stop, watch IsPlaying with GetPropertyChangedSignal.
Looped	Fires on each natural loop. A manual TimePosition jump doesn't count.

Asset permissions

An audio asset only plays in experiences allowed to use it, meaning ones owned by the asset's creator or granted access. If an asset is silent and never becomes IsReady, check its permissions before your wiring.

AudioDeviceInput

A physical microphone as a stream. Voice chat creates one per voice-eligible player under their Player object (see Voice chat), and you can create your own.

Property	Type	Notes
Player	Player	Whose mic this is. Set it on the server so it replicates. Setting it locally only works when the value is Players.LocalPlayer.
Muted	bool	Scriptable mute. Local push-to-talk sets it on the client, and round-based muting sets it on the server.
Active	bool	Read-only to you. Roblox core scripts set it from the player's own mic toggle. Sound flows only when Active and not Muted.
Volume	float	0–3
AccessType	AccessModifierType	Allow means only the listed users can hear this mic. Deny means the listed users cannot. Set it on the server.

Method	Notes
SetUserIdAccessList({userId})	The list that AccessType applies to. Call it from the server, which replicates it.
GetUserIdAccessList()	Read it back.

The access list controls who receives the voice stream at all, in every graph, not just proximity chat. If you put someone on a team-only allow list, they can't hear that mic over your radio either.

Echo cancellation, noise suppression and gain control exist internally but are Roblox-only and not scriptable.

AudioTextToSpeech

Generates speech from text. It works like an AudioPlayer whose "asset" is a sentence.

Property	Type	Notes
Text	string	Up to 300 characters per request. Must follow the Community Standards.
VoiceId	string	See the voice table below.
AutoLocalize	bool	Default true. Translates Text into each player's language and routes it to the matching voice (same gender). Supports Arabic, Chinese (Mandarin), English, French, German, Hindi, Italian, Japanese, Korean, Portuguese and Spanish.
Pitch	float	−12 to +12 semitones
Speed	float	0.5–2 (speaking rate, pitch kept)
PlaybackSpeed	float	0–20 (tape speed, pitch changes)
Volume	float	0–3
Looping, TimePosition, TimeLength		Same as AudioPlayer
IsLoaded	bool	Read-only. Speech is generated and buffered.
IsPlaying	bool	Read-only, but replicates

Method / event	Notes
LoadAsync() → AssetFetchStatus	Generates speech now and yields until it succeeds or fails. Call it ahead of time to avoid a gap before speaking.
Play() / Pause()	They replicate from server to client. There is no Stop: use Pause and set TimePosition = 0.
Unload()	Frees the generated audio.
GetWaveformAsync(timeRange, samples)	Volume samples of the generated speech
Ended, Looped	Same rules as AudioPlayer

Throttles: at most 1 + 6 × concurrent users requests per minute per experience, and 4 requests per second per user on the client. You can buy more through Extended Services.

Voice IDs

ID	Voice	ID	Voice
1	British male	2	British female
3	US male #1	4	US female #1
5	US male #2	6	US female #2
7	Australian male	8	Australian female
9	Retro #1	10	Retro #2
11	Host voice
101	Spanish male	102	Spanish female
201	German male	202	German female
301	Italian male	302	Italian female
401	French male	402	French female
501	Mandarin male	502	Mandarin female
601	Hindi male	602	Hindi female
701	Japanese male	702	Japanese female
801	Arabic male	802	Arabic female
901	Korean male	902	Korean female
1001	Portuguese male	1002	Portuguese female

AudioListener (as a source)

A listener's Output pin carries everything it hears, already spatialized. It is usually wired to an AudioDeviceOutput, but it can also feed effects, an analyzer, or another emitter (a relay, like a baby monitor). See Spatial audio.

VideoPlayer

VideoPlayer is wirable: its soundtrack can be wired into the audio graph like any other source, for example to an AudioEmitter on the screen part. Start videos on the client where you can, because a server-started video is not guaranteed to be loaded or in sync on every client.