using datastore with trainNetwork

2 次查看（过去 30 天）
Noam
Noam 2023-2-7
评论： Noam 2023-2-7
I'm trying to train mp3/wav files using yamnet, I've created an audioDataStore which load the files, then I transform it to get data that suits to Yamnet, and then I try to train it. I get the folowing error:
Invalid training data. Predictors must be a numeric array, a datastore, or a table. For networks with
sequence input, predictors can also be a cell array of sequences.
Can't I use transformedDataStore with trainNetwork?
This is my code:
net = yamnet;
DS = audioDatastore(FolderName, ...
'FileExtensions',{'.wav','.mp3'},"IncludeSubfolders",true,'LabelSource','foldernames');
TR = transform(DS,@(audio,info)preProcess(audio,info),"IncludeInfo",true);
options = trainingOptions("adam");
trainNetwork(net,TR,options)
function [data,info] = preProcess(audio,info)
data = yamnetPreprocess(audio,info.SampleRate);
end
Thanks

请先登录，再回答此问题。

采纳的回答

Aylin
Aylin 2023-2-7
编辑：Aylin 2023-2-7
Hello Noam, maybe the trainNetwork inputs are in the wrong order? The datastore should be first, followed by the layers and the options. Does this work better?
trainNetwork(TR,net,options)
Rylan
  2 个评论
Noam
Noam 2023-2-7
my bad, actually chaniging to:
trainNetwork(TR,net.Layers,options) do the work.
Thanks!
Noam
Noam 2023-2-7
i added a following question:
https://www.mathworks.com/matlabcentral/answers/1908330-input-datastore-returned-more-than-one-observation-per-row-for-network-input-1

请先登录，再进行评论。

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Built-In Training 的更多信息

标签

产品


版本

R2022b

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by