topic: voice
13 modules sharing the voice name family (derived from the tree's prefix discipline).
The 'voice' topic family in the Nishi sovereign ecosystem focuses on creating and evaluating realistic audio synthesis and encoding. It includes modules like nx_voice_clone_synth, which generates new audio, and nx_voice_codec_v2, the core LPC-residual voice codec that enables efficient encoding and decoding of PCM frames. The nx_voice_eval module serves as an objective evaluator, assessing the naturalness of synthesized audio, ensuring the system maintains high-quality voice reproduction across its modular architecture.
auto-narrated by the local model from this topic's module headers; links verified against the wiki index.
narrated overview -- maintained by the narration lane, module links verified against this wiki.
| module | description | lines | funcs |
|---|---|---|---|
| nx_voice_clone_synth.nx | VOICE-CLONE-001 rung 2: CREATE new audio in a | 78 | 2 |
| nx_voice_clone_synth_gate.nx | REFEREE for VOICE-CLONE-001 rung 2 (create). | 155 | 5 |
| nx_voice_codec.nx | NishiVoice v1: the composition primitive that | 353 | 6 |
| nx_voice_codec_tier.nx | tier-adaptive parameter selection. | 109 | 2 |
| nx_voice_codec_v1_adapter.nx | wraps NishiVoice v1 behind nx_codec. | 230 | 5 |
| nx_voice_codec_v2.nx | NishiVoice v2: the REAL LPC-residual voice codec the v1 skeleton described but didn't build. | 135 | 10 |
| nx_voice_codec_v2_gate.nx | proves the v2 voice codec (nx_voice_codec_v2) as a real encode/decode PAIR: PCM frame -> | 65 | 7 |
| nx_voice_eval.nx | THE RULER (voice-realism census V0): a sovereign OBJECTIVE audio-naturalness evaluator so | 145 | 4 |
| nx_voice_fec.nx | forward error correction for the Nishi voice | 192 | 7 |
| nx_voice_frame.nx | Phase 4a foundation: sovereign voice-codec | 190 | 6 |
| nx_voice_read.nx | VOICE-CLONE-001 rung 5-FUSE: a CLONED VOICE READING WORDS. | 82 | 3 |
| nx_voice_read_gate.nx | REFEREE for VOICE-CLONE-001 rung 5-FUSE (voice reads words). | 155 | 8 |
| nx_voice_v2_gate.nx | proves the NishiVoice v2 q/bit path end-to-end: real LPC residual + entropy coding + inverse | 117 | 8 |