虽然参考内容未直接提供基于MFCC和GMM实现说话人识别的Python代码，但可以根据原理构建示例代码： ```python import numpy as np import os import scipy.io.wavfile as wav from python_speech_features import mfcc from sklearn.mixture import GaussianMixture def extract_mfcc(file_path): rate, sig = wav.read(file_path) mfcc_features = mfcc(sig, rate) return mfcc_features def train_gmm(mfcc_features): gmm = GaussianMixture(n_components=16, covariance_type='diag', n_init=3) gmm.fit(mfcc_features) return gmm def predict_speaker(gmm_models, test_mfcc): scores = [] for model in gmm_models: scores.append(model.score(test_mfcc)) speaker_index = np.argmax(scores) return speaker_index # 示例训练数据路径 train_data_path = 'train_data' gmm_models = [] for speaker_folder in os.listdir(train_data_path): speaker_path = os.path.join(train_data_path, speaker_folder) mfcc_features_list = [] for wav_file in os.listdir(speaker_path): wav_path = os.path.join(speaker_path, wav_file) mfcc_features = extract_mfcc(wav_path) mfcc_features_list.extend(mfcc_features) mfcc_features_array = np.array(mfcc_features_list) gmm = train_gmm(mfcc_features_array) gmm_models.append(gmm) # 示例测试数据 test_file = 'test.wav' test_mfcc = extract_mfcc(test_file) predicted_speaker = predict_speaker(gmm_models, test_mfcc) print(f"Predicted speaker index: {predicted_speaker}") ```