tklink的登录做网站,建站是什么专业,html项目模板下载,小程序项目开发报价文章目录作业1#xff1a;机器翻译1. 日期转换1.1 数据集2. 用注意力模型进行机器翻译2.1 注意力机制3. 可视化注意力作业2#xff1a;触发词检测1. 数据合成#xff1a;创建语音数据集1.1 听一下数据1.2 音频转频谱1.3 生成一个训练样本1.4 全部训练集1.5 开发集2. 模型2.1… 文章目录作业1机器翻译1. 日期转换1.1 数据集2. 用注意力模型进行机器翻译2.1 注意力机制3. 可视化注意力作业2触发词检测1. 数据合成创建语音数据集1.1 听一下数据1.2 音频转频谱1.3 生成一个训练样本1.4 全部训练集1.5 开发集2. 模型2.1 建模2.2 训练2.3 测试模型3. 预测3.3 在开发集上测试4. 用自己的样本测试测试题参考博文 笔记W3.序列模型和注意力机制 作业1机器翻译 建立一个神经元机器翻译NMT模型来将人类可读日期25th of June, 2009翻译成机器可读日期“2009—06—25” 将使用注意力模型来实现这一点这是最复杂的 序列到序列 模型之一 注意安装包 pip install Faker2.0.0 pip install babel导入包 from keras.layers import Bidirectional, Concatenate, Permute, Dot, Input, LSTM, Multiply from keras.layers import RepeatVector, Dense, Activation, Lambda from keras.optimizers import Adam from keras.utils import to_categorical from keras.models import load_model, Model import keras.backend as K import numpy as npfrom faker import Faker import random from tqdm import tqdm from babel.dates import format_date from nmt_utils import * import matplotlib.pyplot as plt %matplotlib inline1. 日期转换 模型将输入以各种可能格式书写的日期例如the 29th of August 1958, 03/30/1968, 24 JUNE 1987并将其转换为标准化、机器可读的日期如 1958-08-29, 1968-03-30, 1987-06-24。我们将让模型学习以通用机器可读格式YYYY-MM-DD输出日期 1.1 数据集 1万条数据 m 10000 dataset, human_vocab, machine_vocab, inv_machine_vocab load_dataset(m)打印看看 dataset[:10]输出 [(9 may 1998, 1998-05-09),(10.11.19, 2019-11-10),(9/10/70, 1970-09-10),(saturday april 28 1990, 1990-04-28),(thursday january 26 1995, 1995-01-26),(monday march 7 1983, 1983-03-07),(sunday may 22 1988, 1988-05-22),(08 jul 2008, 2008-07-08),(8 sep 1999, 1999-09-08),(thursday january 1 1981, 1981-01-01)]上面加载了 datasethuman_vocab: 字典 human readable dates an integer-valued indexmachine_vocab: 字典 machine readable dates an integer-valued indexinv_machine_vocab: 字典machine_vocab的反向映射indices characters Tx 30 # 最大输入长度如果大了就截断 Ty 10 # 输出日期长度 YYYY-MM-DD X, Y, Xoh, Yoh preprocess_data(dataset, human_vocab, machine_vocab, Tx, Ty)print(X.shape:, X.shape) print(Y.shape:, Y.shape) print(Xoh.shape:, Xoh.shape) print(Yoh.shape:, Yoh.shape)输出 X.shape: (10000, 30) Y.shape: (10000, 10) Xoh.shape: (10000, 30, 37) # 37 是 len(human_vocab) Yoh.shape: (10000, 10, 11) # 11 是 日期中的字符种类 0-9 和 ‘-’看看数据数据不够长度的会补充 pad所有 x 都是 30 长度 index 52 print(Source date:, dataset[index][0]) print(Target date:, dataset[index][1]) print() print(Source after preprocessing (indices):, X[index]) print(Target after preprocessing (indices):, Y[index]) print() print(Source after preprocessing (one-hot):, Xoh[index]) print(Target after preprocessing (one-hot):, Yoh[index])输出 Source date: saturday october 9 1976 Target date: 1976-10-09Source after preprocessing (indices): [29 13 30 31 28 16 13 34 0 26 15 30 26 14 17 28 0 12 0 4 12 10 9 3636 36 36 36 36 36] Target after preprocessing (indices): [ 2 10 8 7 0 2 1 0 1 10]Source after preprocessing (one-hot): [[0. 0. 0. ... 0. 0. 0.][0. 0. 0. ... 0. 0. 0.][0. 0. 0. ... 0. 0. 0.]...[0. 0. 0. ... 0. 0. 1.][0. 0. 0. ... 0. 0. 1.][0. 0. 0. ... 0. 0. 1.]] Target after preprocessing (one-hot): [[0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.][0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.][0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.][0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.][1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.][0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.][0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.][1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.][0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.][0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]]2. 用注意力模型进行机器翻译 2.1 注意力机制 contextt∑t′0Txαt,t′at′context^{t} \sum_{t 0}^{T_x} \alpha^{t,t}a^{t}contexttt′0∑Tx​​αt,t′at′ 将输入重复几次https://keras.io/zh/layers/core/#repeatvector 输入张量通过 axis 轴串联起来 https://keras.io/zh/layers/merge/#concatenate_1 https://keras.io/zh/layers/wrappers/#bidirectional # Defined shared layers as global variables repeator RepeatVector(Tx) concatenator Concatenate(axis-1) densor1 Dense(10, activation tanh) densor2 Dense(1, activation relu) activator Activation(softmax, nameattention_weights) # We are using a custom softmax(axis 1) loaded in this notebookdotor Dot(axes 1)注意力计算 # GRADED FUNCTION: one_step_attentiondef one_step_attention(a, s_prev):Performs one step of attention: Outputs a context vector computed as a dot product of the attention weightsalphas and the hidden states a of the Bi-LSTM.Arguments:a -- hidden state output of the Bi-LSTM, numpy-array of shape (m, Tx, 2*n_a)s_prev -- previous hidden state of the (post-attention) LSTM, numpy-array of shape (m, n_s)Returns:context -- context vector, input of the next (post-attetion) LSTM cell### START CODE HERE #### Use repeator to repeat s_prev to be of shape (m, Tx, n_s) so that you can concatenate it with all hidden states a (≈ 1 line)s_prev repeator(s_prev)# Use concatenator to concatenate a and s_prev on the last axis (≈ 1 line)concat concatenator(inputs[a, s_prev])# Use densor1 to propagate concat through a small fully-connected neural network to compute the intermediate energies variable e. (≈1 lines)e densor1(concat)# Use densor2 to propagate e through a small fully-connected neural network to compute the energies variable energies. (≈1 lines)energies densor2(e)# Use activator on energies to compute the attention weights alphas (≈ 1 line)alphas activator(energies)# Use dotor together with alphas and a to compute the context vector to be given to the next (post-attention) LSTM-cell (≈ 1 line)context dotor([alphas, a])### END CODE HERE ###return contextn_a 32 n_s 64 post_activation_LSTM_cell LSTM(n_s, return_state True) output_layer Dense(len(machine_vocab), activationsoftmax)# GRADED FUNCTION: modeldef model(Tx, Ty, n_a, n_s, human_vocab_size, machine_vocab_size):Arguments:Tx -- length of the input sequenceTy -- length of the output sequencen_a -- hidden state size of the Bi-LSTMn_s -- hidden state size of the post-attention LSTMhuman_vocab_size -- size of the python dictionary human_vocabmachine_vocab_size -- size of the python dictionary machine_vocabReturns:model -- Keras model instance# Define the inputs of your model with a shape (Tx,)# Define s0 and c0, initial hidden state for the decoder LSTM of shape (n_s,)X Input(shape(Tx, human_vocab_size))s0 Input(shape(n_s,), names0)c0 Input(shape(n_s,), namec0)s s0c c0# Initialize empty list of outputsoutputs []### START CODE HERE #### Step 1: Define your pre-attention Bi-LSTM. Remember to use return_sequencesTrue. (≈ 1 line)a Bidirectional(LSTM(n_a, return_sequencesTrue))(X)# Step 2: Iterate for Ty stepsfor t in range(Ty):# Step 2.A: Perform one step of the attention mechanism to get back the context vector at step t (≈ 1 line)context one_step_attention(a, s)# Step 2.B: Apply the post-attention LSTM cell to the context vector.# Dont forget to pass: initial_state [hidden state, cell state] (≈ 1 line)s, _, c post_activation_LSTM_cell(context, initial_state[s, c])# Step 2.C: Apply Dense layer to the hidden state output of the post-attention LSTM (≈ 1 line)out output_layer(s)# Step 2.D: Append out to the outputs list (≈ 1 line)outputs.append(out)# Step 3: Create model instance taking three inputs and returning the list of outputs. (≈ 1 line)model Model(inputs[X, s0, c0], outputsoutputs)### END CODE HERE ###return model定义模型 model model(Tx, Ty, n_a, n_s, len(human_vocab), len(machine_vocab))定义优化器、配置模型 ### START CODE HERE ### (≈2 lines) opt Adam(learning_rate0.005, beta_10.9, beta_20.999,decay0.01) model.compile(losscategorical_crossentropy,optimizeropt, metrics[accuracy]) ### END CODE HERE ###训练 s0 np.zeros((m, n_s)) c0 np.zeros((m, n_s)) outputs list(Yoh.swapaxes(0,1))model.fit([Xoh, s0, c0], outputs, epochs1, batch_size100)为了节省时间老师准备好了训练好的权值 model.load_weights(models/model.h5)EXAMPLES [5th Otc 2019, 5 April 09, 21th of August 2016, Tue 10 Jul 2007, Saturday May 9 2018, March 3 2001, March 3rd 2001, 1 March 2001] for example in EXAMPLES:source string_to_int(example, Tx, human_vocab)source np.array(list(map(lambda x: to_categorical(x, num_classeslen(human_vocab)), source))).swapaxes(0,1)source source.transpose() #交换两个轴source np.expand_dims(source, axis0) #增加一维轴prediction model.predict([source, s0, c0])prediction np.argmax(prediction, axis -1)output [inv_machine_vocab[int(i)] for i in prediction]print(source:, example)print(output:, .join(output))输出 source: 5th Otc 2019 output: 2019-10-05 source: 5 April 09 output: 2009-04-05 source: 21th of August 2016 output: 2016-08-20 source: Tue 10 Jul 2007 output: 2007-07-10 source: Saturday May 9 2018 output: 2018-05-09 source: March 3 2001 output: 2001-03-03 source: March 3rd 2001 output: 2001-03-03 source: 1 March 2001 output: 2001-03-013. 可视化注意力 attention_map plot_attention_map(model, human_vocab, inv_machine_vocab, Tuesday 09 Oct 1993, num 7, n_s 64)可以看出大部分的注意力用来预测年份 作业2触发词检测 导入包 import numpy as np from pydub import AudioSegment import random import sys import io import os import glob import IPython from td_utils import * %matplotlib inline1. 数据合成创建语音数据集 1.1 听一下数据 有正向音频 activates触发词、负向音频非触发词、背景噪声 IPython.display.Audio(./raw_data/backgrounds/1.wav)1.2 音频转频谱 音频为 44100 Hz 的时长 10秒 x graph_spectrogram(audio_examples/example_train.wav)本作业训练样本时长 10 秒频谱时间步为 5511所以 Tx5511T_x 5511Tx​5511 _, data wavfile.read(audio_examples/example_train.wav) print(Time steps in audio recording before spectrogram, data[:,0].shape) print(Time steps in input after spectrogram, x.shape)输出 Time steps in audio recording before spectrogram (441000,) Time steps in input after spectrogram (101, 5511)定义参数 Tx 5511 # The number of time steps input to the model from the spectrogram n_freq 101 # Number of frequencies input to the model at each time step of the spectrogram Ty 1375 # The number of time steps in the output of our model1.3 生成一个训练样本 随机选择10s 背景噪声随机插入 0-4 段 触发词音频随机插入 0-2 段 非触发词音频 # Load audio segments using pydub activates, negatives, backgrounds load_raw_audio()print(background len: str(len(backgrounds[0]))) # Should be 10,000, since it is a 10 sec clip print(activate[0] len: str(len(activates[0]))) # Maybe around 1000, since an activate audio clip is usually around 1 sec (but varies a lot) print(activate[1] len: str(len(activates[1]))) # Different activate clips can have different lengths 输出 background len: 10000 activate[0] len: 721 activate[1] len: 731获取背景音频中的随机时间段 def get_random_time_segment(segment_ms):Gets a random time segment of duration segment_ms in a 10,000 ms audio clip.Arguments:segment_ms -- the duration of the audio clip in ms (ms stands for milliseconds)Returns:segment_time -- a tuple of (segment_start, segment_end) in mssegment_start np.random.randint(low0, high10000-segment_ms) # Make sure segment doesnt run past the 10sec background segment_end segment_start segment_ms - 1return (segment_start, segment_end)检测插入的音频是否重叠 # GRADED FUNCTION: is_overlappingdef is_overlapping(segment_time, previous_segments):Checks if the time of a segment overlaps with the times of existing segments.Arguments:segment_time -- a tuple of (segment_start, segment_end) for the new segmentprevious_segments -- a list of tuples of (segment_start, segment_end) for the existing segmentsReturns:True if the time segment overlaps with any of the existing segments, False otherwisesegment_start, segment_end segment_time### START CODE HERE ### (≈ 4 line)# Step 1: Initialize overlap as a False flag. (≈ 1 line)overlap False# Step 2: loop over the previous_segments start and end times.# Compare start/end times and set the flag to True if there is an overlap (≈ 3 lines)for previous_start, previous_end in previous_segments:if previous_end segment_start and previous_start segment_end:overlap True### END CODE HERE ###return overlap插入音频 # GRADED FUNCTION: insert_audio_clipdef insert_audio_clip(background, audio_clip, previous_segments):Insert a new audio segment over the background noise at a random time step, ensuring that the audio segment does not overlap with existing segments.Arguments:background -- a 10 second background audio recording. audio_clip -- the audio clip to be inserted/overlaid. previous_segments -- times where audio segments have already been placedReturns:new_background -- the updated background audio# Get the duration of the audio clip in mssegment_ms len(audio_clip)### START CODE HERE ### # Step 1: Use one of the helper functions to pick a random time segment onto which to insert # the new audio clip. (≈ 1 line)segment_time get_random_time_segment(segment_ms)# Step 2: Check if the new segment_time overlaps with one of the previous_segments. If so, keep # picking new segment_time at random until it doesnt overlap. (≈ 2 lines)while is_overlapping(segment_time, previous_segments):segment_time get_random_time_segment(segment_ms)# Step 3: Add the new segment_time to the list of previous_segments (≈ 1 line)previous_segments.append(segment_time)### END CODE HERE #### Step 4: Superpose audio segment and backgroundnew_background background.overlay(audio_clip, position segment_time[0])return new_background, segment_time插入标签 1 # GRADED FUNCTION: insert_onesdef insert_ones(y, segment_end_ms):Update the label vector y. The labels of the 50 output steps strictly after the end of the segment should be set to 1. By strictly we mean that the label of segment_end_y should be 0 while, the50 followinf labels should be ones.Arguments:y -- numpy array of shape (1, Ty), the labels of the training examplesegment_end_ms -- the end time of the segment in msReturns:y -- updated labels# duration of the background (in terms of spectrogram time-steps)segment_end_y int(segment_end_ms * Ty / 10000.0)# Add 1 to the correct index in the background label (y)### START CODE HERE ### (≈ 3 lines)for i in range(segment_end_y1, segment_end_y51):if i Ty:y[0, i] 1### END CODE HERE ###return y合成训练数据 # GRADED FUNCTION: create_training_exampledef create_training_example(background, activates, negatives):Creates a training example with a given background, activates, and negatives.Arguments:background -- a 10 second background audio recordingactivates -- a list of audio segments of the word activatenegatives -- a list of audio segments of random words that are not activateReturns:x -- the spectrogram of the training exampley -- the label at each time step of the spectrogram# Set the random seednp.random.seed(18)# Make background quieterbackground background - 20### START CODE HERE #### Step 1: Initialize y (label vector) of zeros (≈ 1 line)y np.zeros((1, Ty))# Step 2: Initialize segment times as empty list (≈ 1 line)previous_segments []### END CODE HERE #### Select 0-4 random activate audio clips from the entire list of activates recordingsnumber_of_activates np.random.randint(0, 5)random_indices np.random.randint(len(activates), sizenumber_of_activates)random_activates [activates[i] for i in random_indices]### START CODE HERE ### (≈ 3 lines)# Step 3: Loop over randomly selected activate clips and insert in backgroundfor random_activate in random_activates:# Insert the audio clip on the backgroundbackground, segment_time insert_audio_clip(background, random_activate, previous_segments)# Retrieve segment_start and segment_end from segment_timesegment_start, segment_end segment_time# Insert labels in yy insert_ones(y, segment_end)### END CODE HERE #### Select 0-2 random negatives audio recordings from the entire list of negatives recordingsnumber_of_negatives np.random.randint(0, 3)random_indices np.random.randint(len(negatives), sizenumber_of_negatives)random_negatives [negatives[i] for i in random_indices]### START CODE HERE ### (≈ 2 lines)# Step 4: Loop over randomly selected negative clips and insert in backgroundfor random_negative in random_negatives:# Insert the audio clip on the background background, _ insert_audio_clip(background, random_negative, previous_segments)### END CODE HERE #### Standardize the volume of the audio clip background match_target_amplitude(background, -20.0)# Export new training example file_handle background.export(train .wav, formatwav)print(File (train.wav) was saved in your directory.)# Get and plot spectrogram of the new recording (background with superposition of positive and negatives)x graph_spectrogram(train.wav)return x, yx, y create_training_example(backgrounds[0], activates, negatives)plt.plot(y[0])1.4 全部训练集 老师已经处理完了所有数据 # Load preprocessed training examples X np.load(./XY_train/X.npy) Y np.load(./XY_train/Y.npy)1.5 开发集 使用真人录制的音频 # Load preprocessed dev set examples X_dev np.load(./XY_dev/X_dev.npy) Y_dev np.load(./XY_dev/Y_dev.npy)2. 模型 导入包 from keras.callbacks import ModelCheckpoint from keras.models import Model, load_model, Sequential from keras.layers import Dense, Activation, Dropout, Input, Masking, TimeDistributed, LSTM, Conv1D from keras.layers import GRU, Bidirectional, BatchNormalization, Reshape from keras.optimizers import Adam2.1 建模 模型先由一个 1维的卷积 来抽取一些特征还可以加速GRU计算只需要处理 1375 个时间步而不是5511个 注意不要使用双向RNN我们需要检测到触发词后马上输出动作如果使用双向RNN我们需要等待 10s 音频被记录下来再判断 一些 Keras 参考 conv1d https://keras.io/zh/layers/convolutional/#conv1d BN https://keras.io/zh/layers/normalization/#batchnormalization GRU https://keras.io/zh/layers/recurrent/#gru timedistributed https://keras.io/zh/layers/wrappers/#timedistributed # GRADED FUNCTION: modeldef model(input_shape):Function creating the models graph in Keras.Argument:input_shape -- shape of the models input data (using Keras conventions)Returns:model -- Keras model instanceX_input Input(shape input_shape)### START CODE HERE #### Step 1: CONV layer (≈4 lines)X Conv1D(filters196,kernel_size15,strides4)(X_input) # CONV1DX BatchNormalization()(X) # Batch normalizationX Activation(relu)(X) # ReLu activationX Dropout(rate0.8)(X) # dropout (use 0.8)# Step 2: First GRU Layer (≈4 lines)X GRU(128, return_sequencesTrue)(X) # GRU (use 128 units and return the sequences)X Dropout(rate0.8)(X) # dropout (use 0.8)X BatchNormalization()(X) # Batch normalization# Step 3: Second GRU Layer (≈4 lines)X GRU(128, return_sequencesTrue)(X) # GRU (use 128 units and return the sequences)X Dropout(rate0.8)(X) # dropout (use 0.8)X BatchNormalization()(X) # Batch normalizationX Dropout(rate0.8)(X) # dropout (use 0.8)# Step 4: Time-distributed dense layer (≈1 line)X TimeDistributed(Dense(1, activation sigmoid))(X) # time distributed (sigmoid)### END CODE HERE ###model Model(inputs X_input, outputs X)return model model model(input_shape (Tx, n_freq))model.summary()输出 Model: model_1 _________________________________________________________________ Layer (type) Output Shape Param # input_2 (InputLayer) (None, 5511, 101) 0 _________________________________________________________________ conv1d_2 (Conv1D) (None, 1375, 196) 297136 _________________________________________________________________ batch_normalization_2 (Batch (None, 1375, 196) 784 _________________________________________________________________ activation_2 (Activation) (None, 1375, 196) 0 _________________________________________________________________ dropout_2 (Dropout) (None, 1375, 196) 0 _________________________________________________________________ gru_2 (GRU) (None, 1375, 128) 124800 _________________________________________________________________ dropout_3 (Dropout) (None, 1375, 128) 0 _________________________________________________________________ batch_normalization_3 (Batch (None, 1375, 128) 512 _________________________________________________________________ gru_3 (GRU) (None, 1375, 128) 98688 _________________________________________________________________ dropout_4 (Dropout) (None, 1375, 128) 0 _________________________________________________________________ batch_normalization_4 (Batch (None, 1375, 128) 512 _________________________________________________________________ dropout_5 (Dropout) (None, 1375, 128) 0 _________________________________________________________________ time_distributed_1 (TimeDist (None, 1375, 1) 129 Total params: 522,561 Trainable params: 521,657 Non-trainable params: 9042.2 训练 训练很费时在4000个样本上老师已经训练好了该模型 model load_model(./models/tr_model.h5)再用我们的数据集训练1代 opt Adam(lr0.0001, beta_10.9, beta_20.999, decay0.01) model.compile(lossbinary_crossentropy, optimizeropt, metrics[accuracy])model.fit(X, Y, batch_size 5, epochs1)2.3 测试模型 loss, acc model.evaluate(X_dev, Y_dev) print(Dev set accuracy , acc)输出 25/25 [] - 1s 46ms/step Dev set accuracy 0.9427199959754944但是 准确率 在这里不是一个好的衡量标准因为大部分标签都是0都预测为0准确率也会很高应该用 F1值等 3. 预测 def detect_triggerword(filename):plt.subplot(2, 1, 1)x graph_spectrogram(filename)# the spectogram outputs (freqs, Tx) and we want (Tx, freqs) to input into the modelx x.swapaxes(0,1)x np.expand_dims(x, axis0)predictions model.predict(x)plt.subplot(2, 1, 2)plt.plot(predictions[0,:,0])plt.ylabel(probability)plt.show()return predictions一旦估计了在每个输出步骤检测到单词“activate”的概率当概率高于某个阈值时您可以触发“chiming”声音播放。此外在说“activate”之后有很多个 y 值可能接近1但我们只想发出一次蜂鸣音。所以最多每75个输出步骤插入一个蜂鸣音。这将有助于防止我们为“activate”的单个实例插入两个蜂鸣音。这与计算机视觉的非最大值抑制类似 chime_file audio_examples/chime.wav def chime_on_activate(filename, predictions, threshold):audio_clip AudioSegment.from_wav(filename)chime AudioSegment.from_wav(chime_file)Ty predictions.shape[1]# Step 1: Initialize the number of consecutive output steps to 0consecutive_timesteps 0# Step 2: Loop over the output steps in the yfor i in range(Ty):# Step 3: Increment consecutive output stepsconsecutive_timesteps 1# Step 4: If prediction is higher than the threshold and more than 75 consecutive output steps have passedif predictions[0,i,0] threshold and consecutive_timesteps 75:# Step 5: Superpose audio and background using pydubaudio_clip audio_clip.overlay(chime, position ((i / Ty) * audio_clip.duration_seconds)*1000)# Step 6: Reset consecutive output steps to 0consecutive_timesteps 0audio_clip.export(chime_output.wav, formatwav)3.3 在开发集上测试 第一段语音有1个触发 filename ./raw_data/dev/1.wav prediction detect_triggerword(filename) chime_on_activate(filename, prediction, 0.5) IPython.display.Audio(./chime_output.wav)第二段语音有2个触发 4. 用自己的样本测试 # Preprocess the audio to the correct format def preprocess_audio(filename):# Trim or pad audio segment to 10000mspadding AudioSegment.silent(duration10000)segment AudioSegment.from_wav(filename)[:10000]segment padding.overlay(segment)# Set frame rate to 44100segment segment.set_frame_rate(44100)# Export as wavsegment.export(filename, formatwav)your_filename audio_examples/my_audio.wav preprocess_audio(your_filename) IPython.display.Audio(your_filename) # listen to the audio you uploaded chime_threshold 0.5 prediction detect_triggerword(your_filename) chime_on_activate(your_filename, prediction, chime_threshold) IPython.display.Audio(./chime_output.wav)本文地址https://michael.blog.csdn.net/article/details/108933798 我的CSDN博客地址 https://michael.blog.csdn.net/ 长按或扫码关注我的公众号Michael阿明一起加油、一起学习进步