Skip to main content

Consistent duplicated lines in transcripts

Hi, I’m evaluating your API and am finding duplicated transcript lines pretty consistently, sometimes transcribed differently. It seems the repeat might be happening in the input to the transcription model maybe?

Examples:

“(…)

i was surprised too now the most fun part cleaning the mint same oil fresh mint leaves straight in the smell hit the entire room instantly then chili heat the entire room instantly then chili garlic everything back in i heat two minutes (…)”

  • this example also has low quality sentence structuring


“(…) Now our sauce is just equal parts sugar and my dear mum fish sauce, a tablespoon of minced garlic, a splash of water, some chopped Thai chilies to burn your booty hole, Tie chilies to burn your booty hole and lime juice. (…)”

“(…)

Let's start by browning a whole garlic bulb, then add a kilo of pork, mix that around, then 4 tablespoons of vinegar, 2 tablespoons soy sauce, peppercorns, pepper, 1 bay leaf, of vinegar, two tablespoons soy sauce, peppercorns, pepper, one bay leaf, general salt. (..)”

3 comments

Log in to comment and vote

Comments3

  • loic.hovon

    •

    Aug 2

    (from instagram)

    • Rafal

      Team•

      Sep 4

      Hi @loic.hovon this is a very interesting case to investigate, could you please share URLs to posts here or at support@supadata.ai?

      • loic.hovon

        •

        Sep 7

        Sure, sent you an email with more details