ability to identify proper nouns
Jan 6, 2024 15:31:53 GMT 10
Post by account_disabled on Jan 6, 2024 15:31:53 GMT 10
Apossible on a global scale is through automated transcriptions. Manual transcriptions would require considerable time and effort for editors. KQEDs Olson notes that the level of accuracy needs to be high for audio transcriptions especially when it comes to indexing audio news. The advances made so far in speechtotext conversion do not currently meet those standards. Limitations of current speechtotext technology Google tested KQED and KUNGFU.AI applying the latest Speech to Text Tools to a collection of audio news. Limitations were discovered in AIs also known as named entities. Named entities sometimes need context to be understood to be accurately identified.
Something AI doesnt always have. Olson gives an example of KQED audio news that contains speech full of named entities that are contextual to the Bay Area region KQEDs local news audio is rich with references to named entities related to topics people places and organizations that are contextual to the Bay Area region. Speakers use acronyms like CHP for the California Highway Patrol and the peninsula for the area stretching from San Digital Marketing Service Francisco to San Jose. These are more difficult for artificial intelligence to identify. When the named entities are not understood the AI makes its best guess at what was said. However that is an unacceptable solution for web search because an incorrect transcription can change the entire.
Meaning of what was said. Whats Next Work will continue on audio search with plans to make the technology widely accessible when it is developed. David Stoller lead partner for news and publishing at Google says the technology will be shared openly when work on this project is completed. One of the pillars of the Google New Initiative is to incubate new approaches to difficult problems. Once completed this technology and associated best practices will be shared openly greatly amplifying the anticipated impact. Todays machine learning models arent learning from their mistakes says KQEDs Olson which is where humans may need to intervene. The next step is to.
Something AI doesnt always have. Olson gives an example of KQED audio news that contains speech full of named entities that are contextual to the Bay Area region KQEDs local news audio is rich with references to named entities related to topics people places and organizations that are contextual to the Bay Area region. Speakers use acronyms like CHP for the California Highway Patrol and the peninsula for the area stretching from San Digital Marketing Service Francisco to San Jose. These are more difficult for artificial intelligence to identify. When the named entities are not understood the AI makes its best guess at what was said. However that is an unacceptable solution for web search because an incorrect transcription can change the entire.
Meaning of what was said. Whats Next Work will continue on audio search with plans to make the technology widely accessible when it is developed. David Stoller lead partner for news and publishing at Google says the technology will be shared openly when work on this project is completed. One of the pillars of the Google New Initiative is to incubate new approaches to difficult problems. Once completed this technology and associated best practices will be shared openly greatly amplifying the anticipated impact. Todays machine learning models arent learning from their mistakes says KQEDs Olson which is where humans may need to intervene. The next step is to.