Trying whisper with SimpleDirectoryReader

[myrulezzz]

So i am recording an mp3 and trying to return result fro simpledirecoryreader and make it speech. I can see my request transcripted however i have a problem with the reponse. Here is my response code:

Check if the response is not empty

if response: # Extract the top-ranked document as the answer top_doc = response[0].text
# Print the answer print(top_doc) else: print("No matching documents found.")

Print the API response

print(response)

Generate audio from transcript

language = 'en' tts = gTTS(text=top_doc, lang=language)

Save audio to file

filename = "output.mp3" with open(filename, "wb") as f: tts.write_to_fp(f)

Play audio using default media player

playsound("output.mp3") I get this error: raceback (most recent call last): File "/Users/andreasstylianou/Desktop/openai/./whisperdirectory.py", line 75, in top_doc = response[0].text
TypeError: 'Response' object is not subscriptable Any idea how to fix the response format?

[Disiok]

Hey @myrulezzz , how do you get your response object? Maybe taking a look at https://gpt-index.readthedocs.io/en/latest/reference/response.html would help.

Depends on if you want the final textual response (generated by the LLM), or the source nodes, you would need to call different methods.

[myrulezzz]

thanks @Disiok realy helpful and working

[jerryjliu]

sweet! going to close for now