Show HN: Semantic Search in Podcasts Transcriptions

#1

Post by unmonoqueteclea » Sat, Dec 24, 2022, 10:29 AM UTC

I built a podcast search engine that performs a semantic search within thousands of podcasts transcriptions.

Technologies used are: OpenAI Whisper for transcriptions, BERT embeddings and FAISS as a vector database

Show HN: Semantic Search in Podcasts Transcriptions
voilib.com

Re: Show HN: Semantic Search in Podcasts Transcriptions

#2

Post by unmonoqueteclea » Sat, Dec 24, 2022, 10:36 AM UTC

Are you interested in fasting? Ask Voilib about the benefits: https://voilib.com/query?q=What%20are%20the%20benefits%20of%...

Are you worried about inflation? Ask Voilib how long it will last: https://voilib.com/query?q=How%20long%20will%20inflation%20l...

Are you interested in AI? Ask Voilib how it will change the world: https://voilib.com/query?q=How%20is%20AI%20going%20to%20chan...

Re: Show HN: Semantic Search in Podcasts Transcriptions

#3

Post by gbertb » Sat, Dec 24, 2022, 6:00 PM UTC

this looks awesome! Curious what infrastructure or platforms are you using to run the whole thing?

Re: Show HN: Semantic Search in Podcasts Transcriptions

#4

Post by unmonoqueteclea » Sat, Dec 24, 2022, 11:21 PM UTC

gbertb wrote: post #3

this looks awesome! Curious what infrastructure or platforms are you using to run the whole thing?

Thanks!! I have in my living room a quad core i5 server doing transcriptions all day, and then I move them to a low cost quad core VPS in contabo.com running the whole thing!