/// <summary>
/// Check data consistence between script item and segmentation file.
/// </summary>
/// <param name="script">Script file instance.</param>
/// <param name="item">Script item.</param>
/// <param name="fileMap">File list map.</param>
/// <param name="segmentDir">Segment file directory.</param>
/// <param name="errorSet">Data error set found.</param>
/// <param name="phoneBasedSegment">Phone based alignment or unit based alignment.</param>
public static void ValidateDataAlignment(ScriptFile script, ScriptItem item,
FileListMap fileMap, string segmentDir, DataErrorSet errorSet, bool phoneBasedSegment)
{
string segmentFilePath = Path.Combine(segmentDir, fileMap.Map[item.Id] + ".txt");
SegmentFile segmentFile = new SegmentFile();
segmentFile.Load(segmentFilePath);
if (segmentFile.WaveSegments.Count == 0)
{
string message = string.Format(CultureInfo.InvariantCulture,
"There is no valid alignment data into alignment file.");
errorSet.Errors.Add(new DataError(segmentFilePath, message, item.Id));
}
else if (!segmentFile.WaveSegments[segmentFile.WaveSegments.Count - 1].IsSilencePhone)
{
string message = string.Format(CultureInfo.InvariantCulture,
"The alignment file is invalid, for without silence segment at the end.");
errorSet.Errors.Add(new DataError(segmentFilePath, message, item.Id));
}
else if (!phoneBasedSegment && item.Units.Count != segmentFile.NonSilenceWaveSegments.Count)
{
string message = string.Format(CultureInfo.InvariantCulture,
"script units {0} do not match with non-silence segments {1} in segmentation file.",
item.Units.Count, segmentFile.NonSilenceWaveSegments.Count);
errorSet.Errors.Add(new DataError(script.FilePath, message, item.Id));
}
else if (phoneBasedSegment && item.GetPhones().Length != segmentFile.NonSilenceWaveSegments.Count)
{
string message = string.Format(CultureInfo.InvariantCulture,
"script phones {0} do not match with non-silence segments {1} in segmentation file.",
item.GetPhones().Length, segmentFile.NonSilenceWaveSegments.Count);
errorSet.Errors.Add(new DataError(script.FilePath, message, item.Id));
}
else
{
// go through each segments
if (phoneBasedSegment)
{
string[] phones = item.GetPhones();
for (int i = 0; i < segmentFile.NonSilenceWaveSegments.Count; i++)
{
WaveSegment segment = segmentFile.NonSilenceWaveSegments[i];
if (segment.Label != phones[i])
{
string message = string.Format(CultureInfo.InvariantCulture,
"phone [{0}/{1}] at {2} does not match between script and segment.",
WaveSegment.FormatLabel(phones[i]), segment.Label, i);
errorSet.Errors.Add(new DataError(script.FilePath, message, item.Id));
}
}
}
else
{
for (int i = 0; i < segmentFile.NonSilenceWaveSegments.Count; i++)
{
WaveSegment segment = segmentFile.NonSilenceWaveSegments[i];
TtsUnit unit = item.Units[i];
if (segment.Label != WaveSegment.FormatLabel(unit.MetaUnit.Name))
{
string message = string.Format(CultureInfo.InvariantCulture,
"units [{0}/{1}] at {2} do not match between script and segment.",
WaveSegment.FormatLabel(unit.MetaUnit.Name), segment.Label, i);
errorSet.Errors.Add(new DataError(script.FilePath, message, item.Id));
}
}
}
}
}
public static DataError ReadOneScriptItem(ScriptItem scriptItem,
string sentenceLine, string pronunciationLine,
bool withPron, bool withSid, bool validate)
{
if (scriptItem == null)
{
throw new ArgumentNullException("scriptItem");
}
if (string.IsNullOrEmpty(sentenceLine))
{
throw new ArgumentNullException("sentenceLine");
}
sentenceLine = sentenceLine.Trim();
if (withPron)
{
if (string.IsNullOrEmpty(pronunciationLine))
{
throw new ArgumentNullException("pronunciationLine");
}
pronunciationLine = pronunciationLine.Trim();
}
if (withSid)
{
Match m = Regex.Match(sentenceLine, @"^([0-9a-zA-Z]+)[\t ]+(.+)$");
if (!m.Success)
{
string message = string.Format(CultureInfo.InvariantCulture,
"Invalid format, no sentence id for sentence: '{0}', pronunciation: '{1}'.",
sentenceLine, withPron ? pronunciationLine : "null");
return new DataError(message);
}
scriptItem.Id = m.Groups[1].Value;
scriptItem.Sentence = m.Groups[2].Value.Trim();
}
else
{
scriptItem.Sentence = sentenceLine;
}
if (withPron)
{
// Phone set is case insensitive, so convert pronunciation to lower letter.
scriptItem.Pronunciation = pronunciationLine.ToLower(CultureInfo.InvariantCulture);
}
if (validate)
{
Phoneme phoneme = null;
if (scriptItem.Language != Language.Neutral)
{
phoneme = Localor.GetPhoneme(scriptItem.Language, scriptItem.Engine);
}
try
{
// Check all phonemes, currently for DeDE and JaJP only
if (phoneme != null &&
(scriptItem.Language == Language.DeDE || scriptItem.Language == Language.JaJP))
{
string[] phones = scriptItem.GetPhones();
foreach (string phone in phones)
{
phoneme.TtsPhone2Id(phone);
}
}
if (scriptItem.Language != Language.Neutral &&
(scriptItem.NormalWords == null || scriptItem.NormalWords.Count == 0))
{
string message = string.Format(CultureInfo.InvariantCulture,
"No normal word found in the sentence.");
return new DataError("null", message, scriptItem.Id);
}
}
catch (InvalidDataException ide)
{
return new DataError("null", Helper.BuildExceptionMessage(ide), scriptItem.Id);
}
}
return null;
}
