Breaking the language barrier: how do we quickly add multilanguage support in our AI application?
summary
Shows a practical pipeline using Microsoft Cognitive Services plus Google/Amazon API options: Custom Translator is trained with as few as 2,000 sentences, runs training in hours and deploys with one click. Demonstrates language detection, transliteration, speech-to-text and speech translation, with partial/final transcripts and TTS output. Reports a measured improvement from BLEU 22→26 for a custom model vs generic, notes industry-specific models typically score 30–45 while strong human translation hits 50–75. Includes a Python demo of the end-to-end flow.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:05What is NLP
- 15:53How did we reach the current state
- 19:04Applications
- 20:23Cognitive Services
- 22:28Language Models
- 30:35Machine Translation
- 33:42Speech Translation
- 35:52Custom Translation
- 41:20Useful links
related talks
45mTranscend Language Barriers with Custom Machine Learning APIs: um...
47mBuilding smarter apps with Machine Learning, from magic to reality
50mCognitive Services Extravaganza!
20mPolyglot AI: The Role of Natural Language Processing (NLP)
49mMake your applications intelligent (...)
18mAI translation + AI agents = i18n made easy
14mAI for a Billion Voices: Taking AI where there is Maximum Impact
23mMultilingual Natural Language Processing