Solutions
Resources
Neuphonic © 2026
Solutions
Resources
Neuphonic © 2026
Solutions
Resources
Neuphonic © 2026
NeuTTS-2M: On-device text-to-speech in seven languages.
NeuTTS-2M: On-device text-to-speech in seven languages.
NeuTTS-2M: On-device text-to-speech in seven languages.
Generate natural, multilingual speech with a compact model that runs locally on CPU. Build private voice experiences without a cloud dependency.
Generate natural, multilingual speech with a compact model that runs locally on CPU. Build private voice experiences without a cloud dependency.
Generate natural, multilingual speech with a compact model that runs locally on CPU. Build private voice experiences without a cloud dependency.
[
KEY FEATURES
]
Built to deploy on your terms.
Built to deploy on your terms.
Built to deploy
on your terms.
Run production-grade TTS and voice cloning
on-device with a compact, CPU-friendly architecture.
A lightweight neural codec designed to make
training high-quality text-to-speech models simpler and more accessible.
A lightweight neural codec designed to make training high-quality text-to-speech models simpler and more accessible.
Ultra-realistic Voices
Best-in-class for its size. Produces natural human voices.
Ultra-realistic Voices
Best-in-class for its size. Produces natural human voices.
Ultra-realistic Voices
A speech codec that compresses and reconstructs audio with near-inaudible reconstruction loss.
Instant Voice Cloning
Create a new voice from as little as 3 seconds of reference audio.
Instant Voice Cloning
Create a new voice from as little as 3 seconds of reference audio.
On-Device
Deploy in GGML across phones, laptops, Raspberry Pi, and Jetson Orin Nano.
On-Device
Deploy in GGML across phones, laptops, Raspberry Pi, and Jetson Orin Nano.
Simple Architecture
A straightforward LM + codec architecture distilled from a 0.5B backbone.
Simple Architecture
A straightforward LM + codec architecture distilled from a 0.5B backbone.
Runs on CPU
Run in real time on CPU, without requiring a GPU.
Runs on CPU
Run in real time on CPU, without requiring a GPU.
Secure & Private
Keep sensitive voice data on-device, with no cloud dependency.
Secure & Private
Keep sensitive voice data on-device, with no cloud dependency.
Feature
Specification
#Params (Active)
~60M
CPU Threads
2
Audio Generation RTF
2x
Voice Cloning
Yes
Languages
English, Spanish, German, French, Japanese, Korean, Chinese
Feature
Specification
#Params (Active)
~60M
CPU Threads
2
Audio Generation RTF
2x
Voice Cloning
Yes
Languages
English, Spanish, German, French, Japanese, Korean, Chinese
Feature
Specification
#Params (Active)
~60M
CPU Threads
2
Audio Generation RTF
2x
Voice Cloning
Yes
Languages
English, Spanish, German, French, Japanese, Korean, Chinese
Integrate Neutts-2M into your stack.
Integrate Neutts-2M
into your stack.
Integrate Neutts-2M
into your stack.
Bring multilingual, on-device TTS and voice cloning
into your enterprise stack, with private deployment on your terms.
Bring multilingual, on-device TTS
and voice cloning into your enterprise stack,
with private deployment on your terms.
