Jibo-Revival-Group/JiboExperiments
8
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Normalize transcripts and expand speech regressions

Browse Source
This commit is contained in:
Jacob Dubin
2026-05-17 12:39:46 -05:00
parent d8949fcc9a
commit a2aa9df46a
18 changed files with 66058 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

52
OpenJibo/src/Jibo.Cloud/dotnet/src/Jibo.Cloud.Application/Services/JiboInteractionService.cs
View File

@@ -38,13 +38,25 @@ public sealed class JiboInteractionService(
@"\b(?:volume|loudness)\s+(?:2|two|to)\s+(?<value>10|\d|one|two|three|four|five|six|seven|eight|nine|ten)\b",
RegexOptions.IgnoreCase | RegexOptions.CultureInvariant | RegexOptions.Compiled);
private static readonly Regex CommandPhrasePattern = new(
@"[^\w\s]",
RegexOptions.CultureInvariant | RegexOptions.Compiled);
private static readonly Regex CommandWhitespacePattern = new(
@"\s+",
RegexOptions.CultureInvariant | RegexOptions.Compiled);
private static readonly string[] CommandLeadPhrases =
[
"hey jibo",
"hello jibo",
"hi jibo",
"jibo",
"o",
"oh",
"so",
"well",
"um",
"uh",
"hmm",
"erm",
"ah",
"please",
"ok jibo",
"okay jibo"
];
private static readonly Regex AlarmDeletePattern = new(
@"\b(?:cancel|delete|remove|stop|turn\s+off)\s+(?:the\s+)?(?:alarm|along|elo)\b",
@@ -3543,11 +3555,15 @@ public sealed class JiboInteractionService(
return normalized is
"what is the date" or
"what s the date" or
"what's the date" or
"what date is it" or
"today s date" or
"today date" or
"what's today's date" or
"what is today s date" or
"what s today s date" or
"what's today s date" or
"what's todays date" or
"what is todays date" or
"what s todays date";
}
@@ -4311,11 +4327,13 @@ public sealed class JiboInteractionService(
{
var normalized = NormalizeCommandPhrase(loweredTranscript);
return normalized.StartsWith("what is my favorite", StringComparison.Ordinal) ||
normalized.StartsWith("what s my favorite", StringComparison.Ordinal) ||
normalized.StartsWith("what is my favourite", StringComparison.Ordinal) ||
normalized.StartsWith("what s my favourite", StringComparison.Ordinal) ||
normalized.StartsWith("do you remember my favorite", StringComparison.Ordinal) ||
normalized.StartsWith("do you remember my favourite", StringComparison.Ordinal);
normalized.StartsWith("what s my favorite", StringComparison.Ordinal) ||
normalized.StartsWith("what's my favorite", StringComparison.Ordinal) ||
normalized.StartsWith("what is my favourite", StringComparison.Ordinal) ||
normalized.StartsWith("what s my favourite", StringComparison.Ordinal) ||
normalized.StartsWith("what's my favourite", StringComparison.Ordinal) ||
normalized.StartsWith("do you remember my favorite", StringComparison.Ordinal) ||
normalized.StartsWith("do you remember my favourite", StringComparison.Ordinal);
}
private static string? TryExtractPreferenceLookupCategory(string transcript)
@@ -4878,10 +4896,12 @@ public sealed class JiboInteractionService(
private static string NormalizeCommandPhrase(string value)
{
return CommandWhitespacePattern.Replace(
CommandPhrasePattern.Replace(value.Trim().ToLowerInvariant(), " "),
" ")
.Trim();
var normalized = TranscriptTextNormalizer.NormalizeLooseText(value);
if (string.Equals(normalized, "uh huh", StringComparison.Ordinal) ||
normalized.StartsWith("uh huh ", StringComparison.Ordinal))
return normalized;
return TranscriptTextNormalizer.StripLeadingPhrases(normalized, CommandLeadPhrases);
}
private static string? TryNormalizeVolumeLevel(string token)

41
OpenJibo/src/Jibo.Cloud/dotnet/src/Jibo.Cloud.Application/Services/PersonalReportOrchestrator.cs
View File

@@ -41,6 +41,7 @@ internal static class PersonalReportOrchestrator
"yeah",
"yep",
"yup",
"uh huh",
"sure",
"ok",
"okay",
@@ -260,7 +261,7 @@ internal static class PersonalReportOrchestrator
if (toggles.WeatherEnabled)
{
reportSections.Add("First, your weather.");
reportSections.Add("Weather.");
var weatherDecision = await buildWeatherDecisionAsync(turn, "weather", cancellationToken);
reportSections.Add(weatherDecision.ReplyText);
if (IsWeatherErrorReply(weatherDecision.ReplyText)) serviceError = "weather";
@@ -275,13 +276,6 @@ internal static class PersonalReportOrchestrator
catalog.CalendarNothingReplies,
"Looking at your calendar, I don't see anything scheduled today."),
userName));
reportSections.Add(
RenderReportSkillTemplate(
ChooseReportSkillTemplate(
catalog.CalendarOutroReplies,
[],
"And that's it."),
userName));
}
if (toggles.CommuteEnabled)
@@ -302,7 +296,7 @@ internal static class PersonalReportOrchestrator
catalog.NewsCategoryIntroReplies,
"Here's today's news, from the associated press."),
userName));
reportSections.Add(randomizer.Choose(catalog.NewsBriefings));
reportSections.Add(ChooseShortestBriefing(catalog.NewsBriefings));
reportSections.Add(
RenderReportSkillTemplate(
ChooseReportSkillTemplate(
@@ -632,7 +626,8 @@ internal static class PersonalReportOrchestrator
var speakerAwareTemplate = usableTemplates.FirstOrDefault(static template =>
template.Contains("${speaker}", StringComparison.OrdinalIgnoreCase));
return speakerAwareTemplate ?? usableTemplates[0];
return ChooseShortestTemplate(speakerAwareTemplate is not null ? [speakerAwareTemplate] : usableTemplates)
?? fallback;
}
private static string RenderPersonalReportTemplate(string template, string userName)
@@ -649,13 +644,33 @@ internal static class PersonalReportOrchestrator
IReadOnlyList<string> secondaryTemplates,
string fallback)
{
var primary = primaryTemplates.FirstOrDefault(static template => !string.IsNullOrWhiteSpace(template));
var primary = ChooseShortestTemplate(primaryTemplates);
if (!string.IsNullOrWhiteSpace(primary)) return primary!;
var secondary = secondaryTemplates.FirstOrDefault(static template => !string.IsNullOrWhiteSpace(template));
var secondary = ChooseShortestTemplate(secondaryTemplates);
return !string.IsNullOrWhiteSpace(secondary) ? secondary! : fallback;
}
private static string ChooseShortestBriefing(IReadOnlyList<string> briefings)
{
var selected = ChooseShortestTemplate(briefings);
if (string.IsNullOrWhiteSpace(selected)) return string.Empty;
var firstSentence = selected.Split(['.', '!', '?'], 2, StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries)
.FirstOrDefault();
return string.IsNullOrWhiteSpace(firstSentence) ? selected : firstSentence;
}
private static string? ChooseShortestTemplate(IEnumerable<string> templates)
{
var selected = templates
.Where(static template => !string.IsNullOrWhiteSpace(template))
.OrderBy(static template => template.Length)
.FirstOrDefault();
return selected;
}
private static string RenderReportSkillTemplate(string template, string userName)
{
return template
@@ -670,4 +685,4 @@ internal static class PersonalReportOrchestrator
bool CalendarEnabled,
bool CommuteEnabled,
bool NewsEnabled);
}
}

58
OpenJibo/src/Jibo.Cloud/dotnet/src/Jibo.Cloud.Application/Services/TranscriptTextNormalizer.cs Normal file
View File

@@ -0,0 +1,58 @@
using System.Text.RegularExpressions;
namespace Jibo.Cloud.Application.Services;
internal static partial class TranscriptTextNormalizer
{
private static readonly Regex PunctuationToSpaceRegex = new(
@"[^\p{L}\p{N}\s']+",
RegexOptions.CultureInvariant | RegexOptions.Compiled);
private static readonly Regex WhitespaceRegex = new(
@"\s+",
RegexOptions.CultureInvariant | RegexOptions.Compiled);
internal static string NormalizeLooseText(string? value)
{
if (string.IsNullOrWhiteSpace(value)) return string.Empty;
return WhitespaceRegex.Replace(
PunctuationToSpaceRegex.Replace(value.Trim().ToLowerInvariant(), " "),
" ")
.Trim();
}
internal static string StripLeadingPhrases(string value, params string[] phrases)
{
if (string.IsNullOrWhiteSpace(value) || phrases.Length == 0) return value;
var normalized = value;
while (TryStripLeadingPhrase(normalized, phrases, out var trimmed))
normalized = trimmed;
return normalized;
}
private static bool TryStripLeadingPhrase(string normalizedValue, IReadOnlyList<string> phrases, out string trimmed)
{
foreach (var phrase in phrases)
{
if (string.IsNullOrWhiteSpace(phrase)) continue;
if (string.Equals(normalizedValue, phrase, StringComparison.Ordinal))
{
trimmed = string.Empty;
return true;
}
if (normalizedValue.StartsWith($"{phrase} ", StringComparison.Ordinal))
{
trimmed = normalizedValue[(phrase.Length + 1)..].TrimStart();
return true;
}
}
trimmed = normalizedValue;
return false;
}
}

11
OpenJibo/src/Jibo.Cloud/dotnet/src/Jibo.Cloud.Application/Services/WebSocketTurnFinalizationService.cs
View File

@@ -90,6 +90,8 @@ public sealed partial class WebSocketTurnFinalizationService(
private static readonly string[] TranscriptNoisePrefixes =
[
"o",
"oh",
"uh",
"um",
"hmm",
@@ -1627,11 +1629,7 @@ public sealed partial class WebSocketTurnFinalizationService(
private static string NormalizeTranscript(string? transcript)
{
if (string.IsNullOrWhiteSpace(transcript)) return string.Empty;
return TranscriptNormalizationRegex().Replace(transcript.Trim().ToLowerInvariant(), " ")
.Replace(" ", " ", StringComparison.Ordinal)
.Trim();
return TranscriptTextNormalizer.NormalizeLooseText(transcript);
}
private static string? ReadMessageType(TurnContext turn)
@@ -1939,9 +1937,6 @@ public sealed partial class WebSocketTurnFinalizationService(
};
}
[GeneratedRegex(@"[^\w\s]")]
private static partial Regex TranscriptNormalizationRegex();
private enum YesNoReply
{
None = 0,