Speech Recognition: How to Build Speech-to-Text Yourself
summary
Overview of building a Speech-to-Text system for tasks like call-center automation and smart-home voice control, comparing available market options from large IT vendors and smaller companies. Covers practical steps and considerations for implementing an automatic speech recognition (ASR) solution, with references to the speaker’s slides and abstract from HighLoad++ Siberia 2018.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:18Choosing open source engine
- 8:17Training corpus preparation
- 16:26Phonemes and transcription dictionary
- 23:03Hidden Markov models
- 26:56Language models and n-grams
- 33:38System quality evaluation
- 38:13Noise suppression and neural networks
- 43:06Conclusion and future plans
- 47:18Q&A
related talks
46mIntroduction to Speech Recognition Systems Through a Tester’s Eyes
45mDeveloping a Speech to Text component using Azure Logic Apps and Cognitive Services
47mThe Digital Hearth: An In-the-Wild Automated Group Speech to Text System
50mSpeech-to-Text APIs: Their Advantages and Limitations
26mEnabling the next generation in voice technology
36mVoice Enabling Mobile Apps and Custom Hardware Devices
19mThe Advantages of Building Your AI Stack In house
42mdata.bythebay.io: Shubho Sengupta,