NeuTTS Air: Open source Text-To-Speech & Voice Cloning Model

Setting up NeuTTS Air to run completely offline on Mac—no cloud, no API keys, just local voice cloning with 3-10 seconds of reference audio. Testing it with JARVIS, Cogman, Rick Sanchez, and my own voice to see if the claims hold up.
We're using Python 3.12, covering the full Mac setup including eSpeak configuration, walking through the refactored example code for local-only inference, and generating audio with CPU-only processing.

Chapters:
00:00 Intro: NeuTTS Air
01:04 Download NeuTTS & Neucodec Model Weights
02:04 NeuTTS Github Repo
02:39 Model Overview
03:45 NeuTTS Air Local Project Setup
07:57 Debugging Code
09:21 First Test: Cloning JARVIS voice
10:36 Second Test: Cloning Cogman's Voice
11:30 Third Test: Cloning Rick Sanchez Voice Pt. 1
12:17 Fourth Test: Cloning My Voice Pt. 1
13:26 Fifth Test: Cloning My Voice Pt. 2
15:27 Sixth Test: Cloning Rick Sanchez Voice Pt. 2
16:34 Final Thoughts & Mac Users PSA
17:11 Echo Outro


🔗 LINKS:

NeuTTS Air GitHub: https://github.com/neuphonic/neutts-air
NeuTTS Air HuggingFace: https://huggingface.co/neuphonic/neut...
NeuCodec HuggingFace: https://huggingface.co/neuphonic/neuc...


#neuttsair #voicecloning #offlinetts #texttospeech #localai #opensourcetts #huggingface #pythonai #voicesynthesis #cpuinference #neuphonic #speechsynthesis #offlinevoicecloning #ttstutorial #aivoice #pythontts #mactutorial #qwenmodel #charactervoices #localinference