Depending on the speaker ID task, IDOL Speech Server can return the following warnings.
Task | Warning message | Recommended action |
---|---|---|
IvSpkIdTrain
IvSpkIdTrainAudio
|
Specified parameter file does not exist | Check that all the feature files listed in the DataList file
exist.
|
Failed to open ivector parameter file | Check that the read permission is set for all the feature files listed in the DataList file. |
|
Failed to read ivector parameter file header | Check that all the feature files listed in the DataList file are valid IVP files as created by IDOL Speech Server. |
|
NInputs in parameter file not consistent with previous file(s) | Ensure that all input feature files being used to train the templates were trained with the same configuration (for example, using the same iVector base pack). | |
Input label file is longer (501 frames) than audio (500 frames) | Ensure that any label files provided to filter the input features match the length of the audio. | |
IvSpkIdDevel
IvSpkIdDevelAudio
|
Segmentation window(s) set, but disabled as input label file(s) provided | If label files are provided for filtering the input features, do not attempt to specify the segmentation window. |
Could not read file (C:\data\myfile.ivp) - is it a valid ivector parameter (IVP) file? | Check that all the feature files listed in the DataList file are valid IVP files as created by IDOL Speech Server. |
|
Incorrect top result for segment (Smith, expected Brown) | Check the content and quality of the development audio file, as well as the audio files used to train the speaker models. | |
IvSpkIdDevelFinal
|
No development scores found for template [Brown] - threshold not set | Ensure that all speaker templates are included when you run the development tasks (ivDevel or ivDevelAudio ). |
No positive examples for template Brown | Ensure that you provide positive audio examples (data from each speaker) for each speaker template. | |
No negative examples for template Brown | Ensure that you provide negative audio examples (data from speakers other than just the trained speaker) for each speaker template. | |
Brown shows significant overlap between TRUE and FALSE scores | Check the content and quality of the development audio files, as well as the audio files used to train the speaker models. |
|