# Speech to text in NodeRed, undesired outputs and confused on inputs

**URL:** <https://discourse.nodered.org/t/speech-to-text-in-nodered-undesired-outputs-and-confused-on-inputs/93004>\
**Category:** General\
**Created:** [9 November 2024 04:14 UTC](https://discourse.nodered.org/t/speech-to-text-in-nodered-undesired-outputs-and-confused-on-inputs/93004 "2024-11-09T04:14:45Z")\
**Posts on this page:** 6\
**Page:** 1

<div class="post-metadata">

**Author:** ![biglargemassivehuge](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/biglargemassivehuge/32/96231_2.png) [@biglargemassivehuge](https://discourse.nodered.org/u/biglargemassivehuge)\
**Post date:** [9 November 2024 04:14 UTC](https://discourse.nodered.org/t/speech-to-text-in-nodered-undesired-outputs-and-confused-on-inputs/93004/1 "2024-11-09T04:14:45Z")

</div>

Hey, I am pretty new to NodeRed, and I am trying to build in Speech to text functionality. Unfortunately, it seems i have little clue as to the inputs and outputs that I should be using and then getting.

 ![image](https://us1.discourse-cdn.com/flex026/uploads/nodered/original/3X/2/4/24236f86e471c6fa0d3d85de55c053766c60f691.png)  
This is the flow that I have right now, I am converting audio bytes that come in arrays into Base64 strings, as per the documentation of the Google STT node [node-red-contrib-google-stt (node) - Node-RED](https://flows.nodered.org/node/node-red-contrib-google-stt),  
 ![image](https://us1.discourse-cdn.com/flex026/uploads/nodered/original/3X/d/6/d695bc6402b0d566d3ab3f8ed2f36571b363e4f3.png)

I see in other flows that other people are probably using "node-red-contrib-wstt-stream-fixed" for their STT node, but I can't download that one

I am in NodeRed v4.0.2

Thanks for your time and attention guys

---

<div class="post-metadata">

**Author:** ![biglargemassivehuge](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/biglargemassivehuge/32/96231_2.png) [@biglargemassivehuge](https://discourse.nodered.org/u/biglargemassivehuge)\
**Post date:** [9 November 2024 23:15 UTC](https://discourse.nodered.org/t/speech-to-text-in-nodered-undesired-outputs-and-confused-on-inputs/93004/2 "2024-11-09T23:15:50Z")

</div>

nvm, fixed lmao  
idk what changed, but i tried something I did before, which was to change the input type to "Content" and then also change the input format to "Linear16", and then it worked. Wish i knew why it didn't work earlier, but eh

---

<div class="post-metadata">

**Author:** ![biglargemassivehuge](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/biglargemassivehuge/32/96231_2.png) [@biglargemassivehuge](https://discourse.nodered.org/u/biglargemassivehuge)\
**Post date:** [10 November 2024 20:54 UTC](https://discourse.nodered.org/t/speech-to-text-in-nodered-undesired-outputs-and-confused-on-inputs/93004/3 "2024-11-10T20:54:51Z")

</div>

haha, now i REALLY wish i knew why it didn't work earlier, because it stopped working after I changed my NodeRed environment

---

<div class="post-metadata">

**Author:** ![Vilnis\_R](https://avatars.discourse-cdn.com/v4/letter/v/a183cd/32.png) [@Vilnis\_R](https://discourse.nodered.org/u/Vilnis_R)\
**Post date:** [12 November 2024 17:18 UTC](https://discourse.nodered.org/t/speech-to-text-in-nodered-undesired-outputs-and-confused-on-inputs/93004/4 "2024-11-12T17:18:47Z")

</div>

You could try to use Rhasspy that has local speech to text engine and that is not bad at all. I am running Rhaspy on Raspberry along with Node-Red. It recognizes activation keyword (in my case Alexa), then recognizes the intent and then the intent is sent to Node-Red where I am constructing feedback TTS and of course the implementation of the received command..

---

<div class="post-metadata">

**Author:** ![biglargemassivehuge](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/biglargemassivehuge/32/96231_2.png) [@biglargemassivehuge](https://discourse.nodered.org/u/biglargemassivehuge)\
**Post date:** [14 November 2024 11:38 UTC](https://discourse.nodered.org/t/speech-to-text-in-nodered-undesired-outputs-and-confused-on-inputs/93004/5 "2024-11-14T11:38:58Z")

</div>

I considered this too, and I wanted to try this so I spent a few hours trying to figure it out, I was on a bit of a crunch so I gave up. If I ever revisit this project, I'll def give this a more fair shot.

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/flex026/uploads/nodered/original/1X/d073cd938eafa2e558d7c2cd59003b3ef4963033.png) [@system](https://discourse.nodered.org/u/system)\
**Post date:** [28 November 2024 11:39 UTC](https://discourse.nodered.org/t/speech-to-text-in-nodered-undesired-outputs-and-confused-on-inputs/93004/6 "2024-11-28T11:39:10Z")

</div>

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.