Machine translation
Machine translation is the automated computational conversion of text or speech from one natural language into another.
Inside AI
Automatic translation is the computational conversion of content from one human language to another.
- How it is used
- Uses translation to illustrate both cross-language communication and the difficulty of automating seemingly simple human tasks.
- How deep it goes
- Explains contextual and idiomatic failures through examples without detailing translation-system implementation.
- Examples
- Illustrates errors with a French sentence about an avocat and a saying resembling a familiar idiom.
- What you can do after
- Evaluate translation output for contextual and idiomatic errors rather than assuming it is reliable.
Starting Data Analytics with Generative AI and Python
Code conversion is the transformation of source code from one programming language into equivalent code in another language.
- How it is used
- Uses code conversion to adapt analytics workflows for performance or framework compatibility.
- How deep it goes
- Explains potential benefits and risks and illustrates conversion with Python and Java bootstrap code.
- Examples
- Illustrates Python and Java bootstrap confidence interval routines and a potential Python-to-Java or Scala conversion for Hadoop.
- What you can do after
- Evaluate code conversion as a performance or compatibility option while weighing its cost and risk of bugs.
Read how Starting Data Analytics with Generative AI and Python explains machine translation →
Learn Generative AI with PyTorch
The task of automatically translating text from one natural language to another.
- How it is used
- Uses English-to-French translation as a sequence-to-sequence task for building and training an encoder–decoder Transformer.
- How deep it goes
- Builds the Transformer line by line, explains attention and token representations, and proceeds to training and inference.
- Examples
- Uses paired English and French phrases, including “I don’t speak French,” and a dataset of more than 47,000 translation pairs.
- What you can do after
- Build it from scratch by implementing and training a Transformer to translate English phrases into French.
Read how Learn Generative AI with PyTorch explains machine translation →
Inside Deep Learning: Math, Algorithms, Models
- How it is used
- Frames machine translation as a practical application of sequence-to-sequence learning, mapping a source-language sequence into a target-language sequence.
- How deep it goes
- Defines unequal-length mappings and details dataset preparation, while leaving recurrent-network and attention implementation to subsequent steps.
- Examples
- Uses English–French sentence pairs and ambiguous words such as “run” to illustrate translation variation.
- What you can do after
- Build a small English–French translation dataset with paired sentences, vocabulary IDs, boundary tokens, and padded batches.
Read how Inside Deep Learning: Math, Algorithms, Models explains machine translation →
How Large Language Models Work
Machine translation is the automated conversion of text or speech from one human language into another by computational systems.
- How it is used
- Deploys machine translation as an automated aid for repetitive, mildly varying language work.
- How deep it goes
- Explains suitable task patterns and economic implications but stops short of describing translation algorithms or implementation.
- Examples
- Uses repetitive translation work, automated translation workflows, and changing demand for translation services as examples.
- What you can do after
- Apply it via automated tools for suitable translation tasks while evaluating novelty, workflow effects, and continuing human demand.
Read how How Large Language Models Work explains machine translation →
Transfer Learning for Natural Language Processing
Machine translation is the automatic translation of text or speech from one natural language to another.
- How it is used
- Introduces translation as an NLP task and uses it to demonstrate pretrained encoder-decoder models for a low-resource language.
- How deep it goes
- Walks through loading a model, generating and inspecting tokens, while leaving refinement on better data for further work.
- Examples
- Translates “My name is Paul” from English to Twi and inspects the generated tokens.
- What you can do after
- Apply it via the transformers library to generate a baseline translation with a pretrained Marian model.
Read how Transfer Learning for Natural Language Processing explains machine translation →
Data Analysis with LLMs
Translation is the conversion of content from a source language into an equivalent target language.
- How it is used
- Supports spoken communication across languages through a translator that turns speech in one language into speech in another.
- How deep it goes
- Traces transcription, prompt-based text translation, and speech generation through a working Python implementation.
- Examples
- Uses spoken English translated into French and English recast in a Shakespearean style.
- What you can do after
- Operate a speech translator by recording input and supplying a target language or style.
Read how Data Analysis with LLMs explains machine translation →
Introduction to Generative AI, Second Edition
Machine translation is the automated conversion of text or speech from one human language into another.
- How it is used
- Automates translation between languages and serves as a long-standing application of language models.
- How deep it goes
- Explains tokenization, representation, processing, decoding, and attention conceptually without detailing implementation.
- Examples
- Uses an English-to-Spanish sentence and traces the history from programmed rules to statistical and LLM-based translation.
- What you can do after
- Understand how a neural model turns an input sentence into a translation and uses attention to weigh relevant input.
Read how Introduction to Generative AI, Second Edition explains machine translation →
Building Specialized AI Systems
Machine translation is the automated conversion of text or speech from one human language to another.
- How it is used
- Translates business questions into executable Cypher queries, removing a technical barrier to graph-database access.
- How deep it goes
- Explains training pairs, terminology, thresholds, relationships, and joins without detailing model architecture or query-generation implementation.
- Examples
- Uses financial-services questions about high-risk clients, geographic analysis, thresholds, and multi-entity joins.
- What you can do after
- Apply fine-tuning to let analysts formulate plain-language questions and receive accurate, executable database queries.
Read how Building Specialized AI Systems explains machine translation →
Microsoft Azure in Action
Machine translation is the automated conversion of text or speech from one natural language into another.
- How it is used
- Provides multilingual captions or dubbed audio for live events and recorded webinars.
- How deep it goes
- Explains the speech-recognition, translation, and optional speech-synthesis pipeline, and contrasts streaming with batch review.
- Examples
- Uses live Spanish and French captions for an author Q&A and dubbed audio for a recorded panel.
- What you can do after
- Apply it through Azure Speech SDKs or REST APIs to create live captions or batch translated transcripts and audio.
Read how Microsoft Azure in Action explains machine translation →
Data Without Labels
Machine translation is the automated conversion of text or speech from one human language to another.
- How it is used
- Positions machine translation as an NLP application for converting text between languages.
- How deep it goes
- Explains grammatical differences as a challenge and names deep learning as a good fit, but gives no translation-specific implementation.
LLM Evaluation and Alignment, The Foundational Ideas
Machine translation is the automated translation of text or speech from one human language into another.
- How it is used
- Evaluates machine-translation output against human references to support faster iteration and system comparison.
- How deep it goes
- Explains BLEU's lexical-overlap approach and practical role, but does not detail translation-system internals.
- Examples
- Uses BLEU benchmark comparisons, WMT, and low-resource Arabic and Telugu evaluation as examples.
- What you can do after
- Evaluate and compare translation systems with BLEU while treating its score as a limited signal rather than semantic judgment.
Read how LLM Evaluation and Alignment, The Foundational Ideas explains machine translation →
Also covered in
- Human-in-the-Loop Machine Learning: Active learning and annotation for human-centered AI
- Build a Large Language Model (From Scratch)
- Spring AI in Action
- Build a Text-to-Image Generator (from Scratch)
- Deep Learning with R, Third Edition
- Machines that ThinkIn early access (MEAP)
- Hugging Face in Action
- Deep Learning with Python, Third Edition
- The Design of Web APIs, Second Edition
- Transformers in Action
What to know first
- natural language processingNatural language processing is the field of computing concerned with computational methods for analyzing, interpreting, transforming, and generating human language in text or speech.
- language modelA language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- ChatGPTChatGPT is OpenAI’s conversational AI service, which lets users interact with generative AI models and related tools through natural-language prompts.
- generative modelingGenerative modeling is a machine-learning approach that learns patterns or a data distribution to generate new samples resembling the data.
Related concepts
- BLEU scoremeasured by · BLEU is an automatic text-generation evaluation metric that compares candidate text with reference text using modified n-gram precision and a brevity penalty.
- language modelused by · A language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- encoderuses · An encoder is a component or process that transforms input into an encoded or internal representation for subsequent processing, transmission, or decoding.
- annotation workforceused by · A group of people who perform data-annotation work.
- ChatGPTuses · ChatGPT is OpenAI’s conversational AI service, which lets users interact with generative AI models and related tools through natural-language prompts.
Questions about machine translation
What is machine translation?
Machine translation is the automated computational conversion of text or speech from one natural language into another.
Which Manning books explain machine translation?
Inside AI, Starting Data Analytics with Generative AI and Python, Learn Generative AI with PyTorch, Inside Deep Learning: Math, Algorithms, Models, How Large Language Models Work, Transfer Learning for Natural Language Processing, Data Analysis with LLMs, Introduction to Generative AI, Second Edition and 4 more.
What should I know before learning machine translation?
Manning books that teach machine translation usually introduce natural language processing, language model, ChatGPT and generative modeling first.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.











