Yes for the facial rig and for offline MetaHuman Animator, with a few checks before you buy. TikFinity and a long livestream are separate from the rig and need their own setup.
Confirm it is actually a MetaHuman, not a mesh that looks like one. Animator, the face board (smile, surprise, anger, fear), and ARKit lip sync only exist if Sydney is a MetaHuman Character you can open in the MetaHuman Creator plugin and Assemble into a Blueprint. A baked FBX with a custom skeleton will not take Animator audio or the standard face controls, no matter what the description calls the rig. Before purchasing, ask the seller for a shot of it open in MetaHuman Creator (identity / face board), the exact UE version they exported, and whether you get the MetaHuman Character asset rather than only a skeletal mesh. Height and body type do not matter for facial support. Re-assemble after opening it in your editor; a 5.6 preset used in a newer version should be re-assembled, not just dropped in.
Expressions. On a real MetaHuman these are poses on the facial control board / face AnimBP. You do not need an extra plugin for smile, laugh, surprise, anger, fear. Keep the face from dropping to a low LOD during the stream, because lower face LODs simplify or drop the rig. Framing the character large enough, or pinning the face LOD, is the usual fix.
Prerecorded WAV through MetaHuman Animator. That part works in-editor: Animator can solve a facial performance from speech audio and give you an AnimSequence on the MetaHuman face. That is an offline solve, not a live node. For the livestream itself, if the character has to talk to a live mic or TTS, you still need a realtime source of facial curves into the same rig: Live Link Face (iPhone / webcam ARKit), or a runtime audio-to-face tool that outputs ARKit curves (NVIDIA Audio2Face via Live Link is the common one). The standard MetaHuman face already has the ARKit mapping, so those sources connect without a new rig. Animator alone will not lip-sync a live show.
TikFinity gifts. The MetaHuman rig will play any montage or facial pose you trigger. It does not listen to TikTok. TikFinity has to push gift events into Unreal (their own integration / socket; this is not built into the engine), and a Blueprint on your side maps “gift X” to a montage or a face pose. Budget that as custom glue, or a community bridge, not as a MetaHuman feature. Test one gift end-to-end before you design a whole reaction library.
Transitions. Use the default split: body on the body AnimBP, face on the facial AnimBP. Then a seated dance, an idle, and a gift montage can blend on the body (state machine or slots, blend in/out around 0.2–0.4s) while lip sync keeps running on the face. A single full-body animation that also keys the face will clobber the lip sync. Put talk and gift facial reactions on the face layer, and body reactions on a montage slot.
RTX 5070 laptop, 32 GB, four apps at once. One MetaHuman in a simple set is fine on that GPU. The risk is running the editor plus OBS plus TikTok LIVE Studio plus TikFinity for hours, especially on an 8 GB-class laptop VRAM budget. Practical setup: stream a packaged build or Standalone, not the editor (the editor hitches and creeps over a long session). Turn off hardware ray tracing, drop Lumen to High or lower, and prefer baked or simpler lighting if the background allows. Use hair cards or a reduced groom rather than full strands. Your 2K face / 1K body plan is reasonable. Capture with OBS Game Capture and NVENC, and do not also encode a second time in TikTok LIVE Studio if you can hand OBS’s virtual camera or the stream key to a single encoder. 32 GB is enough only if the editor is closed.
Nothing in that list requires a special Sydney-only plugin if the asset really is a standard MetaHuman. The two things that actually block the plan are a fake “MetaHuman rig” (baked mesh) and expecting Animator or TikFinity to work live with no bridge.