What translation memory formats does DeepL support?

DeepL currently works with .TMX translation memory files, the most common translation memory format. You can easily export to .TMX from other file types.

Which languages can DeepL use translation memories for?

DeepL supports translation memories for any combination of English, German, Japanese, Korean, Chinese, French, Spanish, and Italian.

Can I share translation memories with my wider team in DeepL?

Yes. Use style profiles to assign translation memory to all relevant teams and projects.

Can I apply different translation memories for different projects and parts of my business?

Yes. Just add relevant translation memories to the style profiles you assign to projects and business segments.

Cut down on post-editing time with AI-powered translation memory

Give DeepL the same contextual understanding as your language experts by uploading your translation memory. You’ll get translations that are right first time, dramatically reducing post-editing work and freeing up teams to focus on what matters.

Contact sales See pricing

200,000+ businesses and governments trust DeepL’s Language AI

Transform translation workflows

DeepL automatically recalls and applies your translation memory whenever translations are first generated, reflecting all the adjustments you’ve made in the past, and cutting out the need for post-editing work.

Re-use approved translation segments wherever they’re relevant

Customize translation match thresholds to fit your review process and streamline post-editing with full control

Turn every correction into an investment in future quality and cost-savings

Eliminate additional workflows, and the need to apply changes reactively

Apply translation memory everywhere it’s relevant

Exact matches

When there's an exact match, the source segment in your new translation perfectly matches the segment stored in memory—ready to be applied instantly.

In-context exact (ICE) matches

DeepL can detect in-context exact matches—where a translation matches a segment in your translation memory, and reproduces the surrounding context. This cuts out the need for human review altogether.

Dynamic fuzzy matches

When matches are close but not exact, DeepL generates percentage matching scores. Set customizable thresholds to control when a translation memory can be applied automatically, and when you want your team to check first.