github/NarratoAI
1
0
Fork 0
mirror of https://github.com/linyqh/NarratoAI.git synced 2026-02-03 03:38:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

升级几个包；python升级到py11，准备对比优化视频转录；看中英文提示词效果对比；

Browse Source
This commit is contained in:
linyq 2024-10-28 18:32:17 +08:00
parent 2896966359
commit 2ff72e8b06
3 changed files with 33 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.md
View File

@ -130,7 +130,7 @@ docker-compose up
## 开发 💻 ## 开发 💻
1. 安装依赖 1. 安装依赖
```shell ```shell
conda create -n narratoai python=3.10 conda create -n narratoai python=3.11
conda activate narratoai conda activate narratoai
cd narratoai cd narratoai
pip install -r requirements.txt pip install -r requirements.txt

45
app/services/llm.py
View File

@ -14,6 +14,7 @@ from googleapiclient.errors import ResumableUploadError
from google.api_core.exceptions import * from google.api_core.exceptions import *
from google.generativeai.types import * from google.generativeai.types import *
import subprocess import subprocess
from typing import Union, TextIO
from app.config import config from app.config import config
from app.utils.utils import clean_model_output from app.utils.utils import clean_model_output
@ -353,7 +354,7 @@ def _generate_response(prompt: str, llm_provider: str = None) -> str:
return content.replace("\n", "") return content.replace("\n", "")
def _generate_response_video(prompt: str, llm_provider_video: str, video_file: str | File) -> str: def _generate_response_video(prompt: str, llm_provider_video: str, video_file: Union[str, TextIO]) -> str:
""" """
多模态能力大模型 多模态能力大模型
""" """
@ -780,22 +781,28 @@ def screen_matching(huamian: str, wenan: str, llm_provider: str):
if __name__ == "__main__": if __name__ == "__main__":
# 1. 视频转录 # 1. 视频转录
# video_subject = "第二十条之无罪释放" video_subject = "第二十条之无罪释放"
# video_path = "../../resource/videos/test01.mp4" video_path = "/Users/apple/Desktop/home/pipedream_project/downloads/jianzao.mp4"
# language = "zh-CN" language = "zh-CN"
# gemini_video_transcription(video_subject, video_path, language) gemini_video_transcription(
video_name=video_subject,
video_path=video_path,
language=language,
progress_callback=print,
llm_provider_video="gemini"
)
# 2. 解说文案 # # 2. 解说文案
video_path = "/Users/apple/Desktop/home/NarratoAI/resource/videos/1.mp4" # video_path = "/Users/apple/Desktop/home/NarratoAI/resource/videos/1.mp4"
# video_path = "E:\\projects\\NarratoAI\\resource\\videos\\1.mp4" # # video_path = "E:\\projects\\NarratoAI\\resource\\videos\\1.mp4"
video_plot = """ # video_plot = """
李自忠拿着儿子李牧名下的存折，去银行取钱给儿子救命，却被要求证明"你儿子是你儿子"。 # 李自忠拿着儿子李牧名下的存折，去银行取钱给儿子救命，却被要求证明"你儿子是你儿子"。
走投无路时碰到银行被抢劫，劫匪给了他两沓钱救命，李自忠却因此被银行以抢劫罪起诉，并顶格判处20年有期徒刑。 # 走投无路时碰到银行被抢劫，劫匪给了他两沓钱救命，李自忠却因此被银行以抢劫罪起诉，并顶格判处20年有期徒刑。
苏醒后的李牧坚决为父亲做无罪辩护，面对银行的顶级律师团队，他一个法学院大一学生，能否力挽狂澜，创作奇迹？挥法律之利剑 ，持正义之天平！ # 苏醒后的李牧坚决为父亲做无罪辩护，面对银行的顶级律师团队，他一个法学院大一学生，能否力挽狂澜，创作奇迹？挥法律之利剑 ，持正义之天平！
""" # """
res = generate_script(video_path, video_plot, video_name="第二十条之无罪释放") # res = generate_script(video_path, video_plot, video_name="第二十条之无罪释放")
# res = generate_script(video_path, video_plot, video_name="海岸") # # res = generate_script(video_path, video_plot, video_name="海岸")
print("脚本生成成功:\n", res) # print("脚本生成成功:\n", res)
res = clean_model_output(res) # res = clean_model_output(res)
aaa = json.loads(res) # aaa = json.loads(res)
print(json.dumps(aaa, indent=2, ensure_ascii=False)) # print(json.dumps(aaa, indent=2, ensure_ascii=False))

14
requirements.txt
View File

@ -4,22 +4,20 @@ openai~=1.13.3
faster-whisper~=1.0.1 faster-whisper~=1.0.1
edge_tts~=6.1.15 edge_tts~=6.1.15
uvicorn~=0.27.1 uvicorn~=0.27.1
fastapi~=0.110.0 fastapi~=0.115.4
tomli~=2.0.1 tomli~=2.0.1
streamlit~=1.33.0 streamlit~=1.39.0
loguru~=0.7.2 loguru~=0.7.2
aiohttp~=3.9.3 aiohttp~=3.10.10
urllib3~=2.2.1 urllib3~=2.2.1
pillow~=10.3.0 pillow~=10.4.0
pydantic~=2.6.3 pydantic~=2.6.3
g4f~=0.3.0.4 g4f~=0.3.0.4
dashscope~=1.15.0 dashscope~=1.15.0
google.generativeai>=0.8.2 google.generativeai>=0.8.3
python-multipart~=0.0.9 python-multipart~=0.0.9
redis==5.0.3 redis==5.0.3
# if you use pillow~=10.3.0, you will get "PIL.Image' has no attribute 'ANTIALIAS'" error when resize video opencv-python~=4.10.0.84
# please install opencv-python to fix "PIL.Image' has no attribute 'ANTIALIAS'" error
opencv-python~=4.9.0.80
# for azure speech # for azure speech
# https://techcommunity.microsoft.com/t5/ai-azure-ai-services-blog/9-more-realistic-ai-voices-for-conversations-now-generally/ba-p/4099471 # https://techcommunity.microsoft.com/t5/ai-azure-ai-services-blog/9-more-realistic-ai-voices-for-conversations-now-generally/ba-p/4099471
azure-cognitiveservices-speech~=1.37.0 azure-cognitiveservices-speech~=1.37.0