Skip to content
← Models
Open weightsMyShell.ai/10 trustVerified
MeloTTS logo

MeloTTS

MIT-licensed multilingual TTS that runs in real time on CPU.

Visit official page →

About

MeloTTS is a multilingual text-to-speech library from MyShell.ai covering English (several accents), Spanish, French, Chinese, Japanese, and Korean. It is genuinely MIT-licensed for commercial and non-commercial use, runs fast enough for CPU real-time inference, and supports training on custom datasets. The Chinese voice handles mixed Chinese-English text. Quality is a step below the newest neural TTS systems, and some language frontends pull in extra dependencies, but for a permissively licensed workhorse it is hard to beat.

Specs

Context window

Pricing

Free quota

API access

Weights

Open weights

Released

Not disclosed

Benchmarks

Authors claim CPU real-time inference; no standardized benchmark figures published.

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

text-to-speechmultilingual narrationvoice assistantsaccessibility

Trust score — 7/10

/10 trustVerified

Scored 7/10 (verified tier, metrics-v1). Evidence: MIT licensed for commercial and non-commercial use, confirmed on the repo; Covers 6+ languages with multiple English accents; CPU real-time inference; Repo activity slowed after Dec 2024; community forks continue; Quality trails the newest neural TTS; some frontends need extra dependencies. License: MIT per the myshell-ai/MeloTTS repository..

How the trust score is built

Full methodology →
Adoption · stars/downloads, log-scaled vs peers
7.0/10
Maintenance · recency, releases, issue load
5.0/10
Openness · license + docs
9.0/10
Manual review · sanity read + red-flag check
6.0/10
  • · MIT licensed for commercial and non-commercial use, confirmed on the repo
  • · Covers 6+ languages with multiple English accents; CPU real-time inference
  • · Repo activity slowed after Dec 2024; community forks continue
  • · Quality trails the newest neural TTS; some frontends need extra dependencies

Sources