Phrase TMS: a cloud translation platform with the Language Weaver AI engine

Phrase TMS (known as Memsource until 2022) is a cloud translation management system used by corporate localization teams worldwide. For us it is more than another CAT tool: Phrase integrates the neural machine-translation engine Language Weaver, and that combination powers our machine translation post-editing (MTPE) service.

What Phrase TMS does

• A cloud translation workflow. The project, translation memory, glossary, and QA checks live in one environment; the team works from anywhere, and the project manager sees progress in real time.
• Integrated machine translation. Language Weaver generates a subject-aware first draft; the system clearly separates segments taken from translation memory (the machine never touches those) from machine-translated ones that need post-editing.
• Translation memory + glossary. Just as in Trados or memoQ: previously translated sentences are inserted automatically, terminology is highlighted, repetitions are free.
• Hundreds of supported formats — from Office and PDF to JSON, XLIFF, and CMS exports, which makes Phrase a natural fit for website localization and apps.

Why it matters to you

Speed and budget. On large uniform volumes, MTPE in Phrase delivers results faster and ~30% cheaper than classic translation — while every sentence still passes through the hands of a professional translator rather than being published “straight from the machine.”

An honest split between machine and human. We never pass machine output off as human translation or vice versa: the quote shows exactly what goes through MTPE and what a human translates from scratch. For advertising and sensitive legal texts we will recommend the classic process ourselves.

The same quality control. The glossary, translation memory, human editor, and final Xbench check all run regardless of who wrote the draft — a person or Language Weaver.

How we work in Phrase

1. Suitability check — we determine whether this particular text benefits from MTPE; if not, we honestly recommend classic translation.
2. Project profile — we connect your translation memory, glossary, and a tuned Language Weaver engine profile.
3. Machine draft — segments with no memory matches go through MT; anything translated before is filled in from the TM.
4. Human post-editing — a subject-matter linguist brings every sentence to natural phrasing and precise terminology.
5. Control pass — an editorial read plus Phrase’s automated checks before delivery.

Want to compare the cost of classic translation vs MTPE for your project? Send the files — we’ll return both numbers within 15 minutes. Details are on the MTPE service page.