public void ParseConfig(XmlElement config)
{
if (config == null)
{
throw new ArgumentNullException("config");
}
Debug.Assert(!string.IsNullOrEmpty(config.GetAttribute("language")));
Debug.Assert(!string.IsNullOrEmpty(config.GetAttribute("engine")));
_primaryLanguage = Localor.StringToLanguage(
config.GetAttribute("language"));
_engineType = (EngineType)Enum.Parse(typeof(EngineType),
config.GetAttribute("engine"));
XmlElement eleLangData = config.SelectSingleNode("languageData") as XmlElement;
VoiceCreationLanguageData languageData = new VoiceCreationLanguageData();
if (eleLangData != null)
{
languageData.ParseLanguageDataFromXmlElement(true, eleLangData);
languageData.SetLanguageData(_primaryLanguage);
}
else
{
languageData.CartQuestions = config.SelectSingleNode("question/@path").InnerText;
}
_voiceName = config.GetAttribute("voiceName");
_tokenId = config.GetAttribute("tokenId");
_fontPath = config.SelectSingleNode("font/@path").InnerText;
TtsPhoneSet phoneSet = Localor.GetPhoneSet(_primaryLanguage);
XmlScriptValidateSetting validateSetting = new XmlScriptValidateSetting(phoneSet, null);
_scriptFile = XmlScriptFile.LoadWithValidation(config.SelectSingleNode("script/@path").InnerText, validateSetting);
_scriptFile.Remove(ScriptHelper.GetNeedDeleteItemIds(_scriptFile.ErrorSet));
FileMap = new FileListMap();
FileMap.Load(config.SelectSingleNode("filemap/@path").InnerText);
_weightTable = new WeightTable(_primaryLanguage, _engineType);
_weightTable.Load(config.SelectSingleNode("weighttable/@path").InnerText);
_cartTreeManager = new CartTreeManager();
_cartTreeManager.CartTreeDir = config.SelectSingleNode("treedir/@path").InnerText;
if (!Directory.Exists(_cartTreeManager.CartTreeDir))
{
string message = string.Format(CultureInfo.InvariantCulture,
"The treeDir path does not exist at [{0}]",
_cartTreeManager.CartTreeDir);
throw new DirectoryNotFoundException(message);
}
_cartTreeManager.CartQuestionFile = languageData.CartQuestions;
if (!File.Exists(_cartTreeManager.CartQuestionFile))
{
string message = string.Format(CultureInfo.InvariantCulture,
"The tree question file path does not exist at [{0}]",
_cartTreeManager.CartQuestionFile);
throw new DirectoryNotFoundException(message);
}
_cartTreeManager.UnitDescriptFile = config.SelectSingleNode("unitdescript/@path").InnerText;
if (!File.Exists(_cartTreeManager.UnitDescriptFile))
{
string message = string.Format(CultureInfo.InvariantCulture,
"The unit description file path does not exist at [{0}]",
_cartTreeManager.UnitDescriptFile);
throw new DirectoryNotFoundException(message);
}
_unitFeatureFilePath = config.SelectSingleNode("wavesequence/@path").InnerText;
if (!File.Exists(_unitFeatureFilePath))
{
string message = string.Format(CultureInfo.InvariantCulture,
"The wave sequence file path does not exist at [{0}]",
_unitFeatureFilePath);
throw new DirectoryNotFoundException(message);
}
_wave16kDirectories.Clear();
foreach (XmlNode dirNode in config.SelectNodes("wave16k/@path"))
{
string waveDir = dirNode.InnerText.Trim();
if (!Directory.Exists(waveDir))
{
string message = string.Format(CultureInfo.InvariantCulture,
"The wave16k path does not exist at [{0}]",
waveDir);
throw new DirectoryNotFoundException(message);
}
_wave16kDirectories.Add(waveDir);
}
_segmentDirectories.Clear();
foreach (XmlNode dirNode in config.SelectNodes("segment/@path"))
{
string alignmentDir = dirNode.InnerText.Trim();
if (!Directory.Exists(alignmentDir))
{
string message = string.Format(CultureInfo.InvariantCulture,
"The alignment path does not exist at [{0}]",
alignmentDir);
throw new DirectoryNotFoundException(message);
}
_segmentDirectories.Add(alignmentDir);
}
}
/// <summary>
/// Extracts features from the given script.
/// </summary>
/// <param name="script">
/// The xml script file.
/// </param>
/// <param name="fileListMap">
/// The file list map.
/// </param>
/// <param name="alignmentDir">
/// The alignment directory.
/// </param>
/// <param name="waveDir">
/// The wave directory.
/// </param>
/// <returns>
/// The extracted features in training sentence set.
/// </returns>
/// <exception cref="ArgumentNullException">
/// Exception.
/// </exception>
public TrainingSentenceSet Extract(XmlScriptFile script, FileListMap fileListMap, string alignmentDir,
string waveDir)
{
if (script == null)
{
throw new ArgumentNullException("script");
}
if (fileListMap == null)
{
throw new ArgumentNullException("fileListMap");
}
if (alignmentDir == null)
{
throw new ArgumentNullException("alignmentDir");
}
if (waveDir == null)
{
throw new ArgumentNullException("waveDir");
}
TrainingSentenceSet sentenceSet = new TrainingSentenceSet { FileListMap = fileListMap };
List<string> errList = new List<string>();
foreach (string sid in fileListMap.Map.Keys)
{
ScriptItem item = script.ItemDic[sid];
try
{
// Loads the segmentation file.
SegmentFile segmentFile = new SegmentFile();
segmentFile.Load(fileListMap.BuildPath(alignmentDir, sid, "txt"));
// Loads the waveform file to set the end time of the last segmentation.
WaveFile waveFile = new WaveFile();
waveFile.Load(fileListMap.BuildPath(waveDir, sid, FileExtensions.Waveform));
segmentFile.WaveSegments[segmentFile.WaveSegments.Count - 1].EndTime = waveFile.Duration;
// Extracts the single script item.
Sentence sentence = Extract(item, segmentFile);
sentence.TrainingSet = sentenceSet;
sentenceSet.Sentences.Add(sid, sentence);
}
catch (Exception e)
{
if (!(e is InvalidDataException))
{
throw;
}
// Removes the error sentences.
Logger.Log(Helper.BuildExceptionMessage(e));
script.Remove(sid);
errList.Add(sid);
}
}
fileListMap.RemoveItems(errList);
return sentenceSet;
}
/// <summary>
/// Merge scripts in a folder into a script file.
/// Error items are removed from the output file.
/// </summary>
/// <param name="scriptDir">Dir conataining script file.</param>
/// <param name="errorSet">Error set.</param>
/// <param name="resetId">True means resetting id.</param>
/// <param name="validateSetting">Validation setting.</param>
/// <param name="contentController">Contenct controller.</param>
/// <returns>Xml script file.</returns>
public static XmlScriptFile MergeScripts(string scriptDir, ErrorSet errorSet,
bool resetId, XmlScriptValidateSetting validateSetting, object contentController)
{
if (string.IsNullOrEmpty(scriptDir))
{
throw new ArgumentNullException("scriptDir");
}
if (errorSet == null)
{
throw new ArgumentNullException("errorSet");
}
if (validateSetting == null)
{
throw new ArgumentNullException("validateSetting");
}
if (!Directory.Exists(scriptDir))
{
throw new DirectoryNotFoundException(scriptDir);
}
validateSetting.VerifySetting();
XmlScriptValidationScope scope = validateSetting.ValidationScope;
string[] subFiles = Directory.GetFiles(
scriptDir, "*" + XmlScriptFile.Extension, SearchOption.AllDirectories);
XmlScriptFile mergedScript = new XmlScriptFile();
long id = 0;
foreach (string file in subFiles)
{
XmlScriptFile script = new XmlScriptFile();
script.Load(file, contentController);
if (mergedScript.Language == Language.Neutral)
{
mergedScript.Language = script.Language;
}
else if (mergedScript.Language != script.Language)
{
throw new InvalidDataException(Helper.NeutralFormat("Inconsistent langage in {0}", file));
}
if (scope != XmlScriptValidationScope.None)
{
script.PosSet = validateSetting.PosSet;
script.PhoneSet = validateSetting.PhoneSet;
script.Validate(validateSetting);
script.Remove(GetNeedDeleteItemIds(script.ErrorSet));
}
errorSet.Merge(script.ErrorSet);
foreach (ScriptItem item in script.Items)
{
item.Id = resetId ? Helper.NeutralFormat("{0:D10}", ++id) : item.Id;
ErrorSet addErrors = new ErrorSet();
if (!mergedScript.Add(item, addErrors, false))
{
// Added failed
errorSet.Merge(addErrors);
}
}
}
return mergedScript;
}