https://github.com/mikeesto/gemini-transcribe

Transcribe audio and video files with speaker diarization and logically grouped timestamps
https://github.com/mikeesto/gemini-transcribe

gemini-flash speaker-diarization speech-to-text sveltekit transcription

Last synced: 12 days ago
JSON representation

Transcribe audio and video files with speaker diarization and logically grouped timestamps

• Host: GitHub
• URL: https://github.com/mikeesto/gemini-transcribe
• Owner: mikeesto
• Created: 2024-08-15T08:10:48.000Z (3 months ago)
• Default Branch: master
• Last Pushed: 2024-10-30T04:50:43.000Z (14 days ago)
• Last Synced: 2024-10-30T06:26:54.661Z (14 days ago)
• Topics: gemini-flash, speaker-diarization, speech-to-text, sveltekit, transcription
• Language: TypeScript
• Homepage: https://gemini-transcribe.fly.dev/
• Size: 1.68 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Gemini Transcribe

[https://gemini-transcribe.fly.dev/](https://gemini-transcribe.fly.dev/)

A web application for transcribing audio and video files using Google's Gemini Flash model.

Flash is a very interesting model to explore for audio transcription because:

- We can prompt for specific transcription outputs, as it processes both audio and text inputs

- It has built-in speaker diarization

- It can attempt to detect not only words but also silence, sentiment, and sounds beyond human voices

- It can translate the transcription, in particular to languages other than English

Google claims Flash's word error rate is 9.6% in the FLEURS benchmark (September, 2024).