Nith Text-to-Speech

Nith Text-to-Speech#

Nith is a family of text-to-speech models optimized for local, real-time speech synthesis. All processing runs on-device — from MCUs to GPUs — with no cloud dependency.

Nith begins vocalization as soon as the first text buffer is received, delivering natural-sounding speech on resource-constrained edge hardware.

Key Features#

  • Low-latency — Text is processed in small buffers so vocalizations start quickly

  • Private — All speech synthesis runs locally; no text is sent to the cloud

  • Small footprint, high quality — Natural-sounding speech from models that run on edge hardware

  • Custom vocabulary — Specific pronunciation of brand names, technical jargon, and domain- or application-specific words can be added without retraining

  • Multi-language — English, Spanish, and Chinese (Mandarin); additional languages available upon request

Premium Features

  • Custom voices — New TTS voices can be created from a 6-second audio clip

Models#

Select a model based on your language requirements. All models accept text input, output 16kHz mono WAV audio, and have both male and female versions.

nith-5m-live.en nith-5m-live.es nith-5m-live.zh
Languages English ✓ — —
Spanish — ✓ —
Chinese (Mandarin) — — ✓
Specs Architecture Compound CNN + SSM CNN CNN
Quantization Mixed 8/32-bit Mixed 8/32-bit Mixed 8/32-bit

About the Name#

ABR names its model families after rivers. State space models process sequential data as a continuous flow — always moving forward, maintaining state efficiently over time. The Nith model family is named for the river near ABR’s hometown of Waterloo, Ontario, Canada.

Contents