# Reading accuracy

**URL:** <https://community.deeplearning.ai/t/reading-accuracy/723517>\
**Category:** AI Discussions\
**Tags:** ai-discussions\
**Created:** [November 15, 2024, 6:41pm UTC](https://community.deeplearning.ai/t/reading-accuracy/723517 "2024-11-15T18:41:23Z")\
**Posts on this page:** 9\
**Page:** 1

<div class="post-metadata">

**Author:** ![Gabrielamdm](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/gabrielamdm/32/99707_2.png) [@Gabrielamdm](https://community.deeplearning.ai/u/Gabrielamdm)\
**Post date:** [November 15, 2024, 6:41pm UTC](https://community.deeplearning.ai/t/reading-accuracy/723517/1 "2024-11-15T18:41:23Z")

</div>

Hello, I need to evaluate and compare the accuracy of readings using recordings. Is it more advisable to transcribe speech to text and then evaluate with AI or evaluate the audio directly with AI?

---

<div class="post-metadata">

**Author:** ![TMosh](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/tmosh/32/9321_2.png) [@TMosh](https://community.deeplearning.ai/u/TMosh)\
**Post date:** [November 15, 2024, 8:07pm UTC](https://community.deeplearning.ai/t/reading-accuracy/723517/2 "2024-11-15T20:07:55Z")

</div>

What sort of recordings and readings are you referring to?

---

<div class="post-metadata">

**Author:** ![Gabrielamdm](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/gabrielamdm/32/99707_2.png) [@Gabrielamdm](https://community.deeplearning.ai/u/Gabrielamdm)\
**Post date:** [November 15, 2024, 10:12pm UTC](https://community.deeplearning.ai/t/reading-accuracy/723517/3 "2024-11-15T22:12:19Z")

</div>

The reading aloud of a story.

---

<div class="post-metadata">

**Author:** ![TMosh](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/tmosh/32/9321_2.png) [@TMosh](https://community.deeplearning.ai/u/TMosh)\
**Post date:** [November 15, 2024, 10:16pm UTC](https://community.deeplearning.ai/t/reading-accuracy/723517/4 "2024-11-15T22:16:47Z")

</div>

There are lots of speech-to-text tools already.  
Then you could run the text through other AI for analysis.

---

<div class="post-metadata">

**Author:** ![Gabrielamdm](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/gabrielamdm/32/99707_2.png) [@Gabrielamdm](https://community.deeplearning.ai/u/Gabrielamdm)\
**Post date:** [November 15, 2024, 10:18pm UTC](https://community.deeplearning.ai/t/reading-accuracy/723517/5 "2024-11-15T22:18:20Z")

</div>

And what type of AI would be the most recommendable for comparing the accuracy of the readings?

---

<div class="post-metadata">

**Author:** ![TMosh](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/tmosh/32/9321_2.png) [@TMosh](https://community.deeplearning.ai/u/TMosh)\
**Post date:** [November 15, 2024, 11:24pm UTC](https://community.deeplearning.ai/t/reading-accuracy/723517/6 "2024-11-15T23:24:37Z")

</div>

What two things are you comparing?

---

<div class="post-metadata">

**Author:** ![Gabrielamdm](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/gabrielamdm/32/99707_2.png) [@Gabrielamdm](https://community.deeplearning.ai/u/Gabrielamdm)\
**Post date:** [November 17, 2024, 2:21pm UTC](https://community.deeplearning.ai/t/reading-accuracy/723517/7 "2024-11-17T14:21:12Z")

</div>

I´m comparing two recordings of a reading.

---

<div class="post-metadata">

**Author:** ![TMosh](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/tmosh/32/9321_2.png) [@TMosh](https://community.deeplearning.ai/u/TMosh)\
**Post date:** [November 17, 2024, 5:16pm UTC](https://community.deeplearning.ai/t/reading-accuracy/723517/8 "2024-11-17T17:16:04Z")

</div>

I’ll assume that you first use voice-to-text to generate a transcript of the recording.

A quick internet search turned up this method:

> **[can we compare two documents and get the difference using document...](https://learn.microsoft.com/en-us/answers/questions/2032497/can-we-compare-two-documents-and-get-the-differenc)**
>
> I have a user input document which i want to compare with the general template document using document intelligence and get the difference between the two documents.

---

<div class="post-metadata">

**Author:** ![Gabrielamdm](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/gabrielamdm/32/99707_2.png) [@Gabrielamdm](https://community.deeplearning.ai/u/Gabrielamdm)\
**Post date:** [November 18, 2024, 1:59am UTC](https://community.deeplearning.ai/t/reading-accuracy/723517/9 "2024-11-18T01:59:55Z")

</div>

Thank you, I’m going to check it.