FUSE Transcribe™: AI-Powered Speech to Text Audio Transcription
You already surface content like podcasts and video in your FUSE implementation. By default, FUSE will index what’s available — titles, descriptions/abstracts, tags/categories, authors, etc.
FUSE’s speech to text processor, FUSE Transcribe™, can be added on top of any content source collection to also capture the speech that occurs within recorded content.
With the transcripts of that content added to the FUSE index, users’ keyword searches include not only the metadata associated with your recorded content but also the full text that transpires within it. FUSE’s speech recognition uses a deep learning process that will impress you beyond belief. #PowerfulStuff
Conference sessions, board meetings, marketing call recordings — in addition to content such as podcasts — the sky’s the limit with FUSE Transcribe™.
Languages
FUSE Transcribe™ supports over 30 languages. Visit our knowledgebase for supported language details, sorted by language or by country.
Formats
FUSE Transcribe™ supports all major media formats. Visit our knowledgebase for supported file formats.
Custom Vocabulary
What do ATAPI (perhaps pronounced aey-tap-ee), Sundar Pichai (soon-dar pi-chai), and Aquebogue, New York (ack-wuh-bog) have in common?
They’re terms that wouldn’t necessarily be familiar to FUSE’s transcribe engine but that would be important to your users. To account for words, terms, and phrases like this, FUSE works with you to develop a custom vocabulary.
Each custom vocabulary entry has 4 possible components:
Phrase
Sounds Like
Display As
Phonetic Notation
Medical Transcription
If orthopnea, peritoneum, and sagittal are familiar sounding, then FUSE Transcribe™ Healthcare is for you. Designed for organizations with recorded content that skews medicine, healthcare, or academia containing references to terms that relate to, for example, procedures, conditions, diseases, medical/pharmaceutical terminology, or medicine names.
FAQ
Can FUSE surface the transcriptions directly as their own content item?
By default, FUSE associates transcriptions with an existing content item. If there’s a keyword match with a transcript, the content item itself will show in a resultset rather than the transcript directly. However, FUSE has the transcript and can display it in a standalone way if you desire.
Does FUSE filter ‘ums,’ profanity, and the like?
You can eliminate words from being stored in the index that are irrelevant for your users, yes.
How are technical terms or uncommon names handled?
There’s a base vocabulary FUSE uses. However, you can train it to use the nomenclature and syntax familiar to your users (see the Custom Vocabulary and Medical Transcription sections elsewhere on this page).