Help:Automatic transcription: Difference between revisions

From SGUTranscripts
Jump to navigation Jump to search
m (dropping hyphen)
m (a bit more info)
 
Line 1: Line 1:
<big>Voice recognition on SGUTranscripts.org</big>
==Voice recognition on SGUTranscripts.org==
=== Bot Icon: {{bot}} ===
When the AI speech-to-text "robot" icon {{bot}} is used in the episode list, it indicates high quality transcription (2024 standard) from the '''"transcription-bot"''', which does the following:


* performs diarization (identifying speakers)
* maps speakers to the names of the rogues
* segments the transcript (including timestamps)
* uploads the podcast image
* includes links to articles
To report issues or learn more about transcription-bot, visit https://github.com/mheguy/transcription-bot.
=== API Icon: {{a}}===
When the AI speech-to-text icon {{a}} is used in the episode list, it indicates that the episode, or sections of it, were first transcribed by the [https://www.google.com/intl/en/chrome/demos/speech.html Google Web Speech API Demonstration] (or another automatic method).
When the AI speech-to-text icon {{a}} is used in the episode list, it indicates that the episode, or sections of it, were first transcribed by the [https://www.google.com/intl/en/chrome/demos/speech.html Google Web Speech API Demonstration] (or another automatic method).


Although this app works great when one person is speaking clearly, it soon gets things wrong when the Rogues talk over one another, or if someone has an accent other than American. Therefore, when editing automated transcription, or [[Help:How_to_Contribute#Proofreading|proofreading]], you should take this into account.
Although this app works great when one person is speaking clearly, it soon gets things wrong when the Rogues talk over one another, or if someone has an accent other than American.  
<br><br><big>'''''...So, how you can help?!'''''</big><br>
When editing automated transcription, or [[Help:How_to_Contribute#Proofreading|proofreading]], you should take into account the current difficulties that hamper these not-yet-robot-overlords.
[[Category:Help]]
[[Category:Help]]

Latest revision as of 14:41, 7 January 2025

Voice recognition on SGUTranscripts.org[edit]

Bot Icon: Automatic transcription[edit]

When the AI speech-to-text "robot" icon Automatic transcription is used in the episode list, it indicates high quality transcription (2024 standard) from the "transcription-bot", which does the following:

  • performs diarization (identifying speakers)
  • maps speakers to the names of the rogues
  • segments the transcript (including timestamps)
  • uploads the podcast image
  • includes links to articles

To report issues or learn more about transcription-bot, visit https://github.com/mheguy/transcription-bot.

API Icon: Automatic transcription[edit]

When the AI speech-to-text icon Automatic transcription is used in the episode list, it indicates that the episode, or sections of it, were first transcribed by the Google Web Speech API Demonstration (or another automatic method).

Although this app works great when one person is speaking clearly, it soon gets things wrong when the Rogues talk over one another, or if someone has an accent other than American.

...So, how you can help?!
When editing automated transcription, or proofreading, you should take into account the current difficulties that hamper these not-yet-robot-overlords.