Matlab完成数字1-9的语音识别代码

以下是一个基于GMM的Matlab代码示例，用于数字1-9的语音识别：

% Load training data
load('training_data.mat');

% Train GMM model for each digit
num_digits = 9;
models = cell(num_digits, 1);
for i = 1:num_digits
    digit_samples = training_data{i};
    num_samples = size(digit_samples, 1);
    features = zeros(num_samples, num_features);
    for j = 1:num_samples
        sample = digit_samples{j};
        % Preprocess sample
        % ...
        % Extract features from sample
        features(j, :) = extract_features(sample);
    end
    % Train GMM model for digit i
    models{i} = fitgmdist(features, num_gaussians);
end

% Load test data
load('test_data.mat');

% Test GMM model for each test sample
num_test_samples = size(test_data, 1);
results = zeros(num_test_samples, 1);
for i = 1:num_test_samples
    sample = test_data{i};
    % Preprocess sample
    % ...
    % Extract features from sample
    features = extract_features(sample);
    % Test sample against each digit GMM model
    max_prob = -Inf;
    for j = 1:num_digits
        model = models{j};
        prob = pdf(model, features);
        if prob > max_prob
            max_prob = prob;
            results(i) = j;
        end
    end
end

% Evaluate results
correct = 0;
for i = 1:num_test_samples
    if results(i) == test_labels(i)
        correct = correct + 1;
    end
end
accuracy = correct / num_test_samples;
fprintf('Accuracy: %.2f%%\n', accuracy * 100);

其中，`training_data`和`test_data`是已经预处理的训练数据和测试数据，每个元素是一个语音样本。`test_labels`是测试数据对应的标签，表示每个测试样本对应的数字。`num_features`是特征向量的维度，`num_gaussians`是每个数字的GMM模型中高斯分布的个数。`extract_features()`函数用于从语音样本中提取特征，这里可以使用MFCC等方法。