taeha
taeha11mo ago

OpenAI API change not seem to be updated seamlessly, I keep getting the same mp3 file...

I've connected OpenAI's Text generation + TTS + Upload Base64 file + Return so that I can instantly get mp3 file from client-side to play the response. However, when I test it on Buildship or Hoppscotch, whenever I change the prompt or the voice of tts, the output is always the same (which is the first mp3 that I tested). Am I missing something? I've been pushing "Ship" button everytime I make changes. I'm following the exact tutorial "NEW OpenAI Text to Speech API - with No Code". Please help me...
No description
No description
13 Replies
taeha
taeha11mo ago
I've tried to change the file name of the mp3 file but I failed. I need a tip on that one too.. Please please help me is this problem derived from the latency of the buildship cloud?
Gaurav Chadha
Gaurav Chadha11mo ago
Hi I just now tried this, changing the file name worked. I see in the second screenshot, your' Open AI ChatGPT node seems of different version. If you just clone the templeate - Text to Speech and change the file name, you'll get different outputs. I just now tested this and it worked.
No description
No description
taeha
taeha11mo ago
Thank you for helping me out! Even though the ChatGPT node is different version it seems like it works just fine when I call the endpoint from my clientside(React). Hmm. I'll try to use the template. @Gaurav Chadha Also, could you help me out on making Speech to Text (Whisper) version using Buildship???? This would help me understand how to use Buildship very much. Thank you for putting time.
taeha
taeha11mo ago
I'm thinking this is the right way of using it.. But I'm not sure what to do next.
No description
Gaurav Chadha
Gaurav Chadha11mo ago
Yes, it it the write way, you can try testing it, let me know if you encounter any issues. I'll help you thorugh.
taeha
taeha11mo ago
https://storage.googleapis.com/buildship-lvodqd-asia-northeast1/speech.mp3 how can I change the the file name of "speech.mp3" to random named mp3 files??? My current logic does not overwrite the existing "speech.mp3" file so I keep getting the same mp3 result from the past when I ask a new query to GPT. Or is it happening because of other issue?
Gaurav Chadha
Gaurav Chadha11mo ago
Check this screenshot, you can edit file name in field directly -
No description
taeha
taeha11mo ago
But this way I get the mp3 file with the same file name every time I run the API. I want to get the file with a different name every time I run the API. I tried to implement it with the Date() function, but it doesn't output the URL correctly, so I hope you can tell me the proper way to do it.
Gaurav Chadha
Gaurav Chadha11mo ago
No, the file name in the url won't be same, if you look at the test result, you can see my url include speech-2.mp3 and it contains a new output.You can give a try.
taeha
taeha11mo ago
hmm I'm not getting the desired output yet 😦 I only get the expected outcome when I change the value of the return which is Base64 File name. if the File Name is same as before, even though I input different prompt I get the same mp3 file which should be updated
taeha
taeha11mo ago
Also, I get this Error occurring.. It worked fine but now i have this error 😦 What am I doing wrong?
No description
taeha
taeha11mo ago
This is error is gone for some reason Oh.. wait.. I closed the URL and re-opened it. Then it gave me the right result. Refreshing the URL does not seem to load the expected file. Sorry, it wasn't the platform's problem. it was my bad understanding on how the web works. 😛
Gaurav Chadha
Gaurav Chadha11mo ago
Cool 👍