# Normalized samples

**URL:** <https://discourse.mozilla.org/t/normalized-samples/38885>\
**Category:** Common Voice\
**Tags:** feedback\
**Created:** [May 1, 2019, 9:56am UTC](https://discourse.mozilla.org/t/normalized-samples/38885 "2019-05-01T09:56:23Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![mirobiro70](https://sea1.discourse-cdn.com/flex001/user_avatar/discourse.mozilla.org/mirobiro70/32/27722_2.png) [@mirobiro70](https://discourse.mozilla.org/u/mirobiro70)\
**Post date:** [May 1, 2019, 9:56am UTC](https://discourse.mozilla.org/t/normalized-samples/38885/1 "2019-05-01T09:56:23Z")

</div>

After validating many samples, I feel that normalizing the volume of the samples prior to validation would improve the experience for the validator and possibly the accuracy. Just a thought.

---

<div class="post-metadata">

**Author:** ![Michael\_Maggs](https://sea1.discourse-cdn.com/flex001/user_avatar/discourse.mozilla.org/michael_maggs/32/26145_2.png) [@Michael\_Maggs](https://discourse.mozilla.org/u/Michael_Maggs)\
**Post date:** [May 6, 2019, 1:58pm UTC](https://discourse.mozilla.org/t/normalized-samples/38885/2 "2019-05-06T13:58:23Z")

</div>

Hi @ [mirobiro70](https://discourse.mozilla.org/u/mirobiro70), welcome! There is actually  
already an open bug report for this on GitHub, here: [https://github.com/mozilla/voice-web/issues/1498](https://github.com/mozilla/voice-web/issues/1498). Like you, I hope someone is able to implement this.

---

<div class="post-metadata">

**Author:** ![dabinat](https://avatars.discourse-cdn.com/v4/letter/d/8dc957/32.png) [@dabinat](https://discourse.mozilla.org/u/dabinat)\
**Post date:** [May 6, 2019, 4:04pm UTC](https://discourse.mozilla.org/t/normalized-samples/38885/3 "2019-05-06T16:04:45Z")

</div>

Perhaps normalizing is the wrong solution to the problem? Rather than normalizing afterwards, I think it’s better for speakers to record at an appropriate volume in the first place.

The problem is that the only way to tell your microphone is setup correctly is to actually record something in CV and play it back. There needs to be a microphone test function, or at the very least, some kind of UI feedback to show that the recording level is too low.