使用OpenAI Whisper遇导入错误及load_model引用问题求助

问题现象 #

接手的Python音频转写项目使用OpenAI Whisper库时遇到两个问题：

• IDE提示：Cannot find reference 'load_model' in 'whisper.py'
• 运行代码抛出TypeError，报错栈显示导入whisper库时，ctypes.CDLL的参数libc_name为NoneType，无法迭代。

完整报错信息 #

Traceback (most recent call last):
  File "C:\Users\Name\Downloads\Project\Project\Speech to Text\Whisper.py", line 7, in <module>
    import whisper
  File "C:\Users\Name\AppData\Local\Programs\Python\Python310\lib\site-packages\whisper.py", line 69, in <module>
    libc = ctypes.CDLL(libc_name)
  File "C:\Users\Name\AppData\Local\Programs\Python\Python310\lib\ctypes\__init__.py", line 364, in __init__
    if '/' in name or '\\' in name:
TypeError: argument of type 'NoneType' is not iterable

项目代码 #

import os
import tkinter as tk
import tkinter.filedialog as filedialog
import tkinter.messagebox as messagebox
import tkinter.ttk as ttk

import whisper


# Define the function to transcribe the audio file
def transcribe_audio_file():
    # Get the path to the selected audio file
    audio_file_path = filedialog.askopenfilename(
        initialdir=".",
        title="Select audio file",
        filetypes=(
            ("MP3 files", "*.mp3"),
            ("WAV files", "*.wav"),
            ("All files", "*.*")
        )
    )

    # Check if the user cancelled the file dialog
    if not audio_file_path:
        return

    # Get the selected model size
    selected_model = model_var.get()
    model_name = None
    if selected_model == "Tiny":
        model_name = "tiny"
    elif selected_model == "Base":
        model_name = "base"
    elif selected_model == "Small":
        model_name = "small"
    elif selected_model == "Medium":
        model_name = "medium"
    elif selected_model == "Large":
        model_name = "large"

    # Load the Whisper model
    model = whisper.load_model(model_name)

    # Transcribe the audio file
    result = model.transcribe(audio_file_path, fp16=False, language='English')

    # Update the GUI with the transcription
    transcription_text.delete(1.0, tk.END)
    transcription_text.insert(tk.END, result['text'])

    # Save the transcription to a file with the same name as the audio file
    output_file_path = os.path.splitext(audio_file_path)[0] + ".txt"
    with open(output_file_path, "w") as file:
        file.write(result['text'])

    # Show a message box indicating that the transcription was saved
    messagebox.showinfo("Transcription saved",
                        f"The transcription has been saved to file:\n{output_file_path}")

# Function to copy the transcription to clipboard


def copy_to_clipboard():
    transcription = transcription_text.get(1.0, tk.END)
    root.clipboard_clear()
    root.clipboard_append(transcription)


# Create the main window of the GUI
root = tk.Tk()
root.title("Audio Transcription App")

# Create a label and button for selecting the audio file
audio_file_label = tk.Label(root, text="Select an audio file to transcribe:")
audio_file_label.pack()

audio_file_button = tk.Button(
    root, text="Select file", command=transcribe_audio_file)
audio_file_button.pack()

# Create a label and radio buttons for selecting the model size
model_label = tk.Label(root, text="Select a model size:")
model_label.pack()

model_var = tk.StringVar()
model_var.set("Large")

tiny_radio = tk.Radiobutton(
    root, text="Tiny", variable=model_var, value="Tiny")
tiny_radio.pack()

base_radio = tk.Radiobutton(
    root, text="Base", variable=model_var, value="Base")
base_radio.pack()

small_radio = tk.Radiobutton(
    root, text="Small", variable=model_var, value="Small")
small_radio.pack()

medium_radio = tk.Radiobutton(
    root, text="Medium", variable=model_var, value="Medium")
medium_radio.pack()

large_radio = tk.Radiobutton(
    root, text="Large", variable=model_var, value="Large")
large_radio.pack()

# Create a label and text box for displaying the transcription
transcription_label = tk.Label(root, text="Transcription:")
transcription_label.pack()

transcription_text = tk.Text(root, height=10)
transcription_text.pack()

# Create a button to copy the transcription to clipboard
copy_button = ttk.Button(root, text="Copy to Clipboard",
                         command=copy_to_clipboard)
copy_button.pack()

# Start the main loop of the GUI
root.mainloop()

错误原因 #

从报错路径C:\Users\Name\AppData\Local\Programs\Python\Python310\lib\site-packages\whisper.py可以看出，当前安装的不是OpenAI官方的Whisper库，而是一个同名的第三方包。这个包既没有load_model方法，也在初始化时存在libc_name变量未正确赋值的问题，导致两个错误同时出现。

解决方法 #

1. 卸载错误的whisper包 #

执行以下命令卸载当前错误的包：

pip uninstall whisper -y

2. 安装官方OpenAI Whisper库 #

安装官方提供的包：

pip install openai-whisper

3. 安装音频处理依赖（必需） #

Whisper需要ffmpeg处理音频文件，需根据系统安装：

• Windows：下载ffmpeg二进制文件并添加到系统PATH；或使用包管理器choco install ffmpeg
• macOS：brew install ffmpeg
• Linux：sudo apt update && sudo apt install ffmpeg

4. 验证修复 #

运行以下测试代码确认问题解决：

import whisper
model = whisper.load_model("base")
# 替换为你的测试音频路径
result = model.transcribe("test.wav", fp16=False)
print(result["text"])

完成以上步骤后，原项目代码中的import whisper会导入官方库，IDE将不再提示load_model引用错误，运行时的TypeError也会消失。

内容的提问来源于stack exchange，提问作者Thac