Skip to content

personaplex.md

Latest commit

 

History

History
115 lines (91 loc) · 4.13 KB

personaplex.md

File metadata and controls

115 lines (91 loc) · 4.13 KB
Raw

PersonaPlex

PersonaPlex is wired as --family personaplex for speech-to-speech conversation. It consumes user speech, applies a packaged or user-provided voice prompt, and generates an assistant speech response.

Install

python3 tools/model_manager_v2.py install personaplex_7b_v1_q4_k

The Q4_K package is the default. A Q8_0 package is also available:

python3 tools/model_manager_v2.py install personaplex_7b_v1_q8_0

Both packages install into:

models/PersonaPlex-GGUF

Quick Start

Packaged voice prompt:

audiocpp_cli --task s2s --family personaplex \
  --model models/PersonaPlex-GGUF \
  --backend cuda \
  --audio user.wav \
  --text "You are a concise assistant. Answer naturally and briefly." \
  --request-option voice_id=NATF2 \
  --out reply.wav

Raw reference voice:

audiocpp_cli --task s2s --family personaplex \
  --model models/PersonaPlex-GGUF \
  --backend cuda \
  --audio user.wav \
  --voice-ref assets/resources/a.wav \
  --text "You are a calm support specialist. Keep the caller reassured." \
  --out reply_ref.wav

Streaming mode:

audiocpp_cli --task s2s --mode streaming --family personaplex \
  --model models/PersonaPlex-GGUF \
  --backend cuda \
  --audio user.wav \
  --request-option voice_id=NATM1 \
  --text "You are a helpful assistant. Answer the user directly." \
  --out reply_stream.wav

Model

Field Value
Family personaplex
Task s2s
Modes offline, streaming
Model directory models/PersonaPlex-GGUF
Input User speech WAV through --audio
Prompt System/persona prompt through --text or system_prompt
Voice input Packaged voice_id or user WAV through --voice-ref
Language English

If system_prompt is not provided, the runtime uses --text as the system prompt and wraps plain text with the tags expected by the model.

Packaged Voices

The package includes these voice prompt ids:

NATF0 NATF1 NATF2 NATF3
NATM0 NATM1 NATM2 NATM3
VARF0 VARF1 VARF2 VARF3 VARF4
VARM0 VARM1 VARM2 VARM3 VARM4

Use --request-option voice_id=<id> to select one. If omitted, NATF2 is used. If --voice-ref is supplied, the runtime uses that reference audio instead of a packaged prompt.

Options

Option Values Default Meaning
--audio WAV path required User speech input.
--text / --request-option system_prompt=<text> text empty Assistant persona/system prompt.
--request-option voice_id=<id> packaged id NATF2 Packaged PersonaPlex voice prompt id.
--voice-ref WAV path not set User reference voice prompt. Overrides packaged voice_id when present.
--temperature / --request-option temperature=<f> float >= 0 0.8 Audio-token sampling temperature.
--request-option text_temperature=<f> float >= 0 follows temperature Text-token sampling temperature.
--top-k / --request-option top_k=<n> integer >= 0 250 Audio-token top-k sampling limit.
--request-option text_top_k=<n> integer >= 0 follows top_k Text-token top-k sampling limit.
--request-option do_sample=true|false bool true Enable stochastic sampling. Set false for greedy decoding.
--seed / --request-option seed=<n> integer >= 0 42424242 Seed for text and audio token sampling.
--session-option personaplex.graph_arena_mb=<mb> integer MiB 1024 Reusable graph arena size for LM, depformer, and Mimi graphs.
--session-option personaplex.lm_weight_context_mb=<mb> integer MiB 64 Main LM weight metadata arena size.
--session-option personaplex.depformer_weight_context_mb=<mb> integer MiB 64 Depth transformer weight metadata arena size.
--session-option personaplex.mimi_weight_context_mb=<mb> integer MiB 64 Mimi codec weight metadata arena size.
--session-option personaplex.weight_type=<type> native, f32, f16, bf16, q8_0 native LM and Mimi matmul weight storage type when supported.

For GGUF packages, leave weight options at native unless you are deliberately testing conversion or storage behavior.