English ↔ Hebrew Translation

April 6, 2026 · View on GitHub

A curated resource list of translation options supporting English ↔ Hebrew.

Entries are grouped by delivery model (APIs and Local Packages). Some tools appear in both sections — badges indicate type.

Scope

This list focuses on tools that cross the English ↔ Hebrew language boundary. Monolingual Hebrew speech tooling lives in separate resource lists.

In scope

English text ↔ Hebrew text (translation APIs, local packages, LLMs)
Hebrew audio → English text (direct speech translation)
English audio → Hebrew text (direct speech translation)
Hebrew audio ↔ English audio (speech-to-speech, direct or end-to-end)
English text → Hebrew speech and Hebrew text → English speech (translation + TTS pipelines, when offered as an integrated product)
Hebrew-specific support tooling that materially aids translation workflows: nikud/diacritic restoration, transliteration, OCR of Hebrew text, dictionaries

Out of scope

Hebrew → Hebrew STT (monolingual Hebrew speech-to-text), even though it can be chained with a second-stage translator. See the separate Hebrew STT resource list.
Hebrew → Hebrew TTS (monolingual Hebrew text-to-speech). See the separate Hebrew TTS resource list.
General-purpose multilingual ASR/TTS models that do not specifically expose an EN↔HE translation path.
Translation between Hebrew and languages other than English.

Legend

Type badges:

Per-entry checklist:

Directions: [ ] EN → HE / [ ] HE → EN
Translation type: NMT / SMT / LLM / Hybrid
Enhanced translation: reflows paragraphs, removes redundancy, improves readability beyond raw MT (often an LLM trait)

APIs

Google Cloud Translation

Link: https://cloud.google.com/translate
Directions: [x] EN → HE [x] HE → EN
Translation type: NMT (Neural Machine Translation)
Enhanced translation: [ ] (raw MT; Advanced edition supports glossaries / custom models)
Notes: Mature Hebrew support, pay-as-you-go.

DeepL

Link: https://www.deepl.com/
Directions: [x] EN → HE [x] HE → EN (Hebrew added to DeepL's supported languages — verify current status)
Translation type: NMT
Enhanced translation: [~] (DeepL Write offers rewriting, but not for all languages)
Notes: Often rated highest quality for European languages.

Microsoft Translator (Azure AI Translator)

Link: https://learn.microsoft.com/azure/ai-services/translator/
Directions: [x] EN → HE [x] HE → EN
Translation type: NMT
Enhanced translation: [ ]
Notes: Free tier available, document translation supported.

Lokalise

Link: https://lokalise.com/
Directions: [x] EN → HE [x] HE → EN
Translation type: Hybrid (TMS aggregating Google/DeepL/Microsoft + human workflow)
Enhanced translation: [x] (via AI features and human review)
Notes: Localization platform rather than raw MT engine.

Lingvanex

Link: https://lingvanex.com/services/hebrew-translation-api/
Directions: [x] EN → HE [x] HE → EN
Translation type: NMT
Enhanced translation: [ ]

TranslateX

Link: https://translatex.com/?sl=iw&tl=en
Directions: [x] EN → HE [x] HE → EN
Translation type: NMT (aggregator)

LibreTranslate (hosted)

Link: https://libretranslate.com/
Directions: [x] EN → HE [x] HE → EN (verify model coverage)
Translation type: NMT (Argos Translate / OpenNMT)
Enhanced translation: [ ]
Notes: Open source, self-hostable. See Local Packages section.

Local Packages

LibreTranslate (self-hosted)

Link: https://github.com/LibreTranslate/LibreTranslate
Directions: [x] EN → HE [x] HE → EN
Translation type: NMT (Argos Translate models)
Enhanced translation: [ ]
Notes: Docker-friendly, fully offline-capable.

Argos Translate

Link: https://github.com/argosopentech/argos-translate
Directions: [x] EN → HE [x] HE → EN (via community packages)
Translation type: NMT (OpenNMT)
Enhanced translation: [ ]

Helsinki-NLP / Opus-MT

Link: https://huggingface.co/Helsinki-NLP
Directions: [x] EN → HE (opus-mt-en-he) [x] HE → EN (opus-mt-he-en)
Translation type: NMT (MarianMT)
Enhanced translation: [ ]
Notes: Run via transformers, fully local.

NLLB-200 (Meta)

Link: https://huggingface.co/facebook/nllb-200-distilled-600M
Directions: [x] EN → HE [x] HE → EN
Translation type: NMT (200-language multilingual model)
Enhanced translation: [ ]

LLMs (general-purpose, used for translation)

LLMs are listed separately because they typically perform enhanced translation by default — reflowing paragraphs, removing redundancy, fixing grammar, and adapting tone.

OpenAI (GPT-4o / GPT-4.1 / o-series)

Directions: [x] EN → HE [x] HE → EN
Enhanced translation: [x]

Anthropic Claude

Directions: [x] EN → HE [x] HE → EN
Enhanced translation: [x]
Notes: Strong at preserving tone and structure in long-form Hebrew.

Google Gemini

Directions: [x] EN → HE [x] HE → EN
Enhanced translation: [x]

Local LLMs (Llama 3.x, Qwen, Mistral, etc. via Ollama)

Link: https://ollama.com/
Directions: [x] EN → HE [x] HE → EN (quality varies by model size)
Enhanced translation: [x]
Notes: Larger models (70B+) handle Hebrew noticeably better than small ones.

Speech

Tools that handle audio in the translation pipeline. Broken down by modality:

Text → Speech translation (EN text → HE speech, HE text → EN speech)
Speech → Text translation (HE speech → EN text, EN speech → HE text)
Speech → Speech translation (direct, both directions)

Note: most pipelines today are still ASR + MT + TTS chained. "Direct" entries are end-to-end models.

EN text → HE speech (translation + TTS)

Typically achieved by combining a translation API with a Hebrew TTS voice.

Google Cloud — Translation API + Cloud TTS (Hebrew voices available). https://cloud.google.com/text-to-speech
Microsoft Azure — Translator + Azure Speech (Hebrew neural voices). https://learn.microsoft.com/azure/ai-services/speech-service/
ElevenLabs — Multilingual TTS with Hebrew support. https://elevenlabs.io/

HE text → EN speech

Google Cloud / Azure / ElevenLabs — same stacks as above, reversed.

HE speech → EN text (speech translation)

Soniox — Hebrew ASR with built-in translation to English. https://soniox.com/speech-to-text/hebrew
OpenAI Whisper — translate task transcribes any source language directly to English. Hebrew supported. https://github.com/openai/whisper
Google Cloud Speech-to-Text — Hebrew ASR; pair with Translation API.
Azure Speech Translation — End-to-end speech translation, Hebrew supported as source.

EN speech → HE text

Whisper + MT — Whisper transcribes EN, then any translation API renders HE.
Azure Speech Translation — supports EN → HE as a target.
Google Cloud Speech-to-Text + Translation.

Speech → Speech (direct or near-direct)

Meta SeamlessM4T / Seamless — Direct speech-to-speech translation across many languages including Hebrew (verify HE coverage per model variant). https://huggingface.co/facebook/seamless-m4t-v2-large
Azure Speech Translation — Real-time S2S via chained ASR + MT + TTS, both directions with Hebrew.
Google Translate (mobile "Conversation" mode) — Consumer-grade S2S, both directions.

Hebrew-Specific Models

ashercn97/english-hebrew-translation

Link: https://huggingface.co/ashercn97/english-hebrew-translation
Directions: [x] EN → HE [ ] HE → EN
Translation type: NMT (fine-tuned)

Nikud / Diacritic Restoration

Tools that add vowel points (niqqud) to unpointed Hebrew text — useful as a pre/post-processing step for translation, TTS, and readability.

Dicta Nakdan

Link: https://nakdan.dicta.org.il/
Notes: Web tool and API from Dicta. Considered the gold standard for automatic Hebrew vocalization.

Nakdimon

Link: https://github.com/elazarg/nakdimon
Notes: Open-source neural nikud restoration.

Unikud

Link: https://github.com/morrisalp/unikud
Notes: Transformer-based nikud restoration.

baravninaor/punctuation-restoration-deberta-alephbert-hebrew

Link: https://huggingface.co/baravninaor/punctuation-restoration-deberta-alephbert-hebrew
Notes: Restores punctuation in Hebrew text — complements nikud restoration.

Transliteration

hebrew-transliteration (charlesLoder)

Link: https://github.com/charlesLoder/hebrew-transliteration
Notes: JS/TS library for transliterating Hebrew (including Biblical) to Latin script. See also https://github.com/charlesLoder/hebrewTransliteration

OCR

sivan22/testing-trOCR-hebrew-handwritten

Link: https://huggingface.co/sivan22/testing-trOCR-hebrew-handwritten
Notes: TrOCR fine-tune for handwritten Hebrew.

Dicta

Link: https://dicta.org.il/
Notes: Suite of Hebrew NLP tools — nikud, morphological analysis, OCR, translation, and more. Many are free to use.

Milon

Link: https://github.com/gregarkhipov/milon
Notes: Hebrew dictionary utility.

BlitzAI

Link: https://github.com/hoodini/blitzai
Notes: AI utility (verify scope/relevance to Hebrew translation).

Magento Hebrew Translation

Link: https://github.com/itaymesh/magento-hebrew-translation
Notes: Hebrew localization pack for Magento.

MCP Servers

Model Context Protocol servers exposing translation tooling to AI agents.

DeepL MCP Server

Link: https://github.com/DeepLcom/deepl-mcp-server
Notes: Official DeepL MCP server. Hebrew availability depends on DeepL's current language support.

loco-mcp

Link: https://github.com/valentinperignon/loco-mcp
Notes: MCP server for Loco localization platform.

Localization Platforms

Translation Management Systems (TMS) and localization APIs that handle EN↔HE as part of broader i18n workflows.

SimpleLocalize

Link: https://simplelocalize.io/localization-api/
Notes: Localization API with translation memory and machine translation integrations.

Lokalise

See entry under APIs.

Document & File Translation

Tools for translating specific file formats (preserving structure/layout).

ppt-translator

Link: https://github.com/daekeun-ml/ppt-translator
Notes: Translates PowerPoint files while preserving formatting.

Contributing

PRs welcome. For each entry please include:

Link
Type badges (API / Local, LLM / Translation API)
Direction checkboxes (EN→HE, HE→EN)
Translation type (NMT / SMT / LLM / Hybrid)
Whether it supports enhanced translation (paragraph reflow, redundancy removal, readability improvements)
Pricing model and notes on Hebrew quality