Component	Architecture	Size
Temporal Transformer	32-layer, 4096d, 32 heads (7B params)	~6.5 GB (8-bit)
Depformer	6-layer, 1024d, 16 heads, per-codebook weights	~1.3 GB (8-bit)
Mimi Codec	SEANet encoder/decoder + 8L transformer + 16 RVQ codebooks	~370 MB (fp16)
Embeddings	Text + 16 audio embeddings + output heads	~940 MB (fp16)
Total		~9.1 GB

Usage

Guide: <a href="https://soniqo.audio/guides/respond" target="_blank" rel="noopener noreferrer" class="text-blue-400 hover:text-blue-300 underline underline-offset-2 transition-colors">soniqo.audio/guides/respond
Docs: <a href="https://soniqo.audio" target="_blank" rel="noopener noreferrer" class="text-blue-400 hover:text-blue-300 underline underline-offset-2 transition-colors">soniqo.audio
GitHub: <a href="https://github.com/soniqo/speech-swift" target="_blank" rel="noopener noreferrer" class="text-blue-400 hover:text-blue-300 underline underline-offset-2 transition-colors">soniqo/speech-swift

swift

let model = try await PersonaPlexModel.fromPretrained(
    modelId: "aufklarer/PersonaPlex-7B-MLX-8bit"
)
let response = model.respond(audio: samples, voice: .NATF0, steps: 100)

audio personaplex input.wav --model aufklarer/PersonaPlex-7B-MLX-8bit -o output.wav

Variants

Variant	Quantization	Size	Model ID
4-bit	4-bit	~4.9 GB	aufklarer/PersonaPlex-7B-MLX-4bit
8-bit	8-bit	~9.1 GB	aufklarer/PersonaPlex-7B-MLX-8bit

Voices

18 voice presets available: NATF0-3, NATM0-3, VARF0-4, VARM0-4