Demo Error in PresentAgent Colab Demo

[mathida520]

Hello I try to run the demo in Colab. I followed the instructions and successfully initialized the models. However, in PPT generation, it couldn’t generate any PPT given PDF, and for video generation, the video it generated didn’t have any audio for the transcript. I have attached the videos it generated, and you can find they are both silent.

generated_video_20250829_022822.2.mp4

766.mov

Could you please give any hint to help figure these issues? Thanks a lot!

[shi0712]

For ppt generation failed, you need to provide detail error from the log on colab. For video generation, you need to ensure there has speech text in the remark of each slide

[mathida520]

Hi @shi0712 thanks for your help! following your instruction I add speech text to the remark of each slide, and try again but still failed to have audio in the video. I have attached my slides with speech text and the generated video for it. Could you please have a look again?

Thanks a lot!

LLM-based AI Study Helper 2.pptx
https://github.com/user-attachments/assets/d4d47949-494a-42c9-ad6c-cf50be7c0559

[shi0712]

I find the error is due to update python version of colab, previouly we use python 3.11, so the MegaTTS installation may fail. A solution is remove the line pip install pynini==2.1.5 and replace pip install WeTextProcessing==1.0.4.1 with the following line

[mathida520]

Hi @shi0712 thanks for your help! Yes following your instructions, video generation works now. For ppt generation, the error is:

2025-08-30 01:42:49,878 - WARNING - llms.py:305 - Error in AsyncLLM call: Error code: 429 - {'error': {'message': 'You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.', 'type': 'insufficient_quota', 'param': None, 'code': 'insufficient_quota'}}

However, my GPT API has sufficient quota and didn’t use up. Could you please figure it out? Thanks a lot!

[shi0712]

Can you use python script to call LLM with your key?

[mathida520]

Hi @shi0712 my key is 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. I got it from https://platform.openai.com/api-keys and I used it to successfully initialize the model and run video generation in Colab, so I think it works.

[shi0712]

from openai import OpenAI
client = OpenAI(
api_key="",
)
response = client.chat.completions.create(
model="gpt-4o",
messages=[
{"role": "user", "content": "Hello!"}
]
)
print(response.choices[0].message.content)

You can use this code replacing your api key to test whether it works, you may run !pip install openai on colab

[internetofaiagent]

Hi @shi0712 thanks for reply and I run your code to try again. The result is still RateLimitError: Error code: 429 - {'error': {'message': 'You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.', 'type': 'insufficient_quota', 'param': None, 'code': 'insufficient_quota'}}. Does it mean my free api key doesn't work, and I need to pay for more credits to enable my api key?

[shi0712]

Yes.