window环境 如何替换 听 说 呀，库已经安装好了

[weigtman]

No description provided.

[Is-Jiejie]

请问你在windows环境下如何安装的mlx库呀，是在anaconda中安装的吗。我在pycharm中使用python3.8，除了mlx其余库都可以安装，但是mlx始终无法安装

[Yeeler]

To install from PyPI you must meet the following requirements:

• Using an M series chip (Apple silicon)
• Using a native Python >= 3.8
• macOS >= 13.3

MLX is only available on devices running macOS >= 13.3 It is highly recommended to use macOS 14 (Sonoma)
https://ml-explore.github.io/mlx/build/html/install.html

[Yeeler]

@linyiLYi Is there an alternative lib we can use? I don't have a up-to-date MacOS machine.

[owlsan49]

mlx是mac专用的机器学习库，windows版得改代码

[onedollor]

可以考虑用openai的whisper替换掉whisper目录.

几行code改动,还有貌似torch在cuda 12.4上还有问题，只好用nightly的.

pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu121
torch-2.4.0.dev20240416%2Bcu121-cp39-cp39-win_amd64.whl
tensor([0.], device='cuda:0')
True

还需要ffmpeg,安装可以按照ffmpeg

whisper_model = whisper.load_model("large", download_root=f"{MODEL_ROOT}/openai/whisper")

user_input = whisper_model.transcribe("recordings/output.wav")["text"]

C:\Python\Python39\python.exe I:/AI/code/voice-assistant-main/main.py
torch.cuda.is_available()=>True torch.zeros(1).cuda()=>tensor([0.], device='cuda:0')
好测试再一次测试
Listening...

Error in text-to-speech: [WinError 2] The system cannot find the file specified

speech 部分用以下替换(仅仅用于测试，不连贯，也听起来很傻)

from gtts import gTTS
from pygame import mixer
mixer.init()
import tempfile

    except Exception as e:
        print(f"Error in text-to-speech: {e}")
        self.speak(text, lang="zh-cn")

def speak(self, text, lang):
    with tempfile.NamedTemporaryFile(delete=True) as fp:
        tts = gTTS(text=text, lang=lang)
        tts.save('{}.mp3'.format(fp.name))
        mixer.music.load('{}.mp3'.format(fp.name))
        mixer.music.play(1)