Adaptiven A1-Kurs und statistisches Aussprachetraining integrieren
Direkte DeepSeek- und Mistral-Anbindung mit versionierten Aktionsprofilen, Budgetreservierung, Tokenabrechnung und administrativen Qualitätsprüfungen ersetzen die Flowise-Abhängigkeit im regulären Betrieb. 840 geprüfte A1-Aufgaben mit geschützter Testauswertung und konkreten offenen Arbeitsaufträgen ergänzen den Kurs. Inhalts- und Sprachbewertung bleiben von Azure-Aussprachewerten getrennt; historische Aufgaben und Antworten bleiben erhalten. Echte Wort- und Phonemmessungen werden mit Herkunft, Aktualität und Nachweisdichte gebündelt. Go bestimmt Trainingsziele und Wiederholungen, das Modell formuliert dazu passende Nachsprechübungen. Profilansicht und Auswahlgründe werden angezeigt. Validierung: Go-Race-Tests mit lokaler PostgreSQL-Integration, go vet, Web-Build, Audit aller 840 Aufgaben und drei Browser-Smokes mit lokalen Modell-/Speech-Fixtures bestanden. Keine Produktionsänderungen oder Live-Aufrufe.
This commit is contained in:
@@ -53,19 +53,32 @@ func (s *AzureSpeechService) AssessPronunciation(ctx context.Context, input Pron
|
||||
return domain.PronunciationResult{}, errors.New("target text is required")
|
||||
}
|
||||
if s.cfg.AzureMock || s.cfg.AzureSpeechKey == "" || s.cfg.AzureRegion == "" {
|
||||
return mockPronunciation(input.TargetText), nil
|
||||
result := mockPronunciation(input.TargetText)
|
||||
result.Locale = input.Locale
|
||||
result.Method = "scripted"
|
||||
result.MeasurementVersion = "demo-v1"
|
||||
return result, nil
|
||||
}
|
||||
|
||||
endpoint := fmt.Sprintf("https://%s.stt.speech.microsoft.com/speech/recognition/conversation/cognitiveservices/v1?language=%s",
|
||||
s.cfg.AzureRegion, url.QueryEscape(valueOr(input.Locale, "en-US")))
|
||||
|
||||
assessmentConfig := map[string]any{
|
||||
"ReferenceText": input.TargetText,
|
||||
"GradingSystem": "HundredMark",
|
||||
"Granularity": "Phoneme",
|
||||
"Dimension": "Comprehensive",
|
||||
"EnableMiscue": true,
|
||||
"PhonemeAlphabet": "IPA",
|
||||
"ReferenceText": input.TargetText,
|
||||
"GradingSystem": "HundredMark",
|
||||
"Granularity": "Phoneme",
|
||||
"Dimension": "Comprehensive",
|
||||
"EnableMiscue": true,
|
||||
}
|
||||
alphabet := ""
|
||||
switch valueOr(input.Locale, "en-US") {
|
||||
case "en-US":
|
||||
alphabet = "IPA"
|
||||
case "zh-CN":
|
||||
alphabet = "SAPI"
|
||||
}
|
||||
if alphabet != "" {
|
||||
assessmentConfig["PhonemeAlphabet"] = alphabet
|
||||
}
|
||||
assessmentJSON, _ := json.Marshal(assessmentConfig)
|
||||
|
||||
@@ -92,7 +105,12 @@ func (s *AzureSpeechService) AssessPronunciation(ctx context.Context, input Pron
|
||||
s.logger.Warn("azure pronunciation failed", "status", resp.StatusCode, "body", string(body))
|
||||
return domain.PronunciationResult{}, fmt.Errorf("azure pronunciation status %d", resp.StatusCode)
|
||||
}
|
||||
return parseAzurePronunciation(body)
|
||||
result, err := parseAzurePronunciation(body)
|
||||
result.Locale = valueOr(input.Locale, "en-US")
|
||||
result.Alphabet = alphabet
|
||||
result.Method = "scripted"
|
||||
result.MeasurementVersion = "azure-rest-v1"
|
||||
return result, err
|
||||
}
|
||||
|
||||
func (s *AzureSpeechService) Synthesize(ctx context.Context, locale, voice, text string) ([]byte, string, error) {
|
||||
@@ -178,12 +196,12 @@ type azurePronunciationResponse struct {
|
||||
ProsodyScore float64 `json:"ProsodyScore"`
|
||||
} `json:"PronunciationAssessment"`
|
||||
Words []struct {
|
||||
Word string `json:"Word"`
|
||||
ErrorType string `json:"ErrorType"`
|
||||
AccuracyScore float64 `json:"AccuracyScore"`
|
||||
Word string `json:"Word"`
|
||||
ErrorType string `json:"ErrorType"`
|
||||
AccuracyScore *float64 `json:"AccuracyScore"`
|
||||
PronunciationAssessment struct {
|
||||
AccuracyScore float64 `json:"AccuracyScore"`
|
||||
ErrorType string `json:"ErrorType"`
|
||||
AccuracyScore *float64 `json:"AccuracyScore"`
|
||||
ErrorType string `json:"ErrorType"`
|
||||
} `json:"PronunciationAssessment"`
|
||||
Phonemes []struct {
|
||||
Phoneme string `json:"Phoneme"`
|
||||
@@ -204,6 +222,9 @@ func parseAzurePronunciation(body []byte) (domain.PronunciationResult, error) {
|
||||
if len(raw.NBest) == 0 {
|
||||
return domain.PronunciationResult{}, fmt.Errorf("azure response has no NBest result")
|
||||
}
|
||||
if raw.RecognitionStatus != "" && raw.RecognitionStatus != "Success" {
|
||||
return domain.PronunciationResult{}, fmt.Errorf("Azure recognition did not succeed: %s", raw.RecognitionStatus)
|
||||
}
|
||||
best := raw.NBest[0]
|
||||
result := domain.PronunciationResult{
|
||||
RecognizedText: firstNonEmpty(best.Display, raw.DisplayText),
|
||||
@@ -217,7 +238,7 @@ func parseAzurePronunciation(body []byte) (domain.PronunciationResult, error) {
|
||||
for _, word := range best.Words {
|
||||
phonemes := make([]string, 0, len(word.Phonemes))
|
||||
phonemeScores := make([]domain.PhonemeScore, 0, len(word.Phonemes))
|
||||
for _, phoneme := range word.Phonemes {
|
||||
for position, phoneme := range word.Phonemes {
|
||||
if phoneme.Phoneme != "" {
|
||||
phonemes = append(phonemes, phoneme.Phoneme)
|
||||
accuracy := phoneme.PronunciationAssessment.AccuracyScore
|
||||
@@ -226,19 +247,28 @@ func parseAzurePronunciation(body []byte) (domain.PronunciationResult, error) {
|
||||
}
|
||||
if accuracy != nil {
|
||||
phonemeScores = append(phonemeScores, domain.PhonemeScore{
|
||||
Phoneme: phoneme.Phoneme, Accuracy: *accuracy, Highlight: highlightFor(*accuracy),
|
||||
Phoneme: phoneme.Phoneme, Accuracy: *accuracy, Highlight: highlightFor(*accuracy), Position: &position,
|
||||
})
|
||||
}
|
||||
}
|
||||
}
|
||||
score := firstScore(word.AccuracyScore, word.PronunciationAssessment.AccuracyScore)
|
||||
measured := word.PronunciationAssessment.AccuracyScore
|
||||
if measured == nil {
|
||||
measured = word.AccuracyScore
|
||||
}
|
||||
hasScore := measured != nil
|
||||
score := 0.0
|
||||
if measured != nil {
|
||||
score = *measured
|
||||
}
|
||||
result.WordScores = append(result.WordScores, domain.WordScore{
|
||||
Word: word.Word,
|
||||
Accuracy: score,
|
||||
ErrorType: firstNonEmpty(word.PronunciationAssessment.ErrorType, word.ErrorType),
|
||||
Phonemes: phonemes,
|
||||
PhonemeScores: phonemeScores,
|
||||
Highlight: highlightFor(score),
|
||||
Word: word.Word,
|
||||
AccuracyMeasured: &hasScore,
|
||||
Accuracy: score,
|
||||
ErrorType: firstNonEmpty(word.PronunciationAssessment.ErrorType, word.ErrorType),
|
||||
Phonemes: phonemes,
|
||||
PhonemeScores: phonemeScores,
|
||||
Highlight: highlightFor(score),
|
||||
})
|
||||
}
|
||||
return result, nil
|
||||
@@ -332,3 +362,47 @@ func valueOr(value, fallback string) string {
|
||||
}
|
||||
return value
|
||||
}
|
||||
|
||||
// Transcribe is separate from reference-based assessment: free speech must not
|
||||
// be forced to match an invented model answer. No mock transcript is fabricated.
|
||||
func (s *AzureSpeechService) Transcribe(ctx context.Context, audio []byte, contentType, locale string) (string, error) {
|
||||
if s.cfg.AzureMock || s.cfg.AzureSpeechKey == "" || s.cfg.AzureRegion == "" {
|
||||
return "", errors.New("free speech transcription requires Azure Speech")
|
||||
}
|
||||
endpoint := fmt.Sprintf("https://%s.stt.speech.microsoft.com/speech/recognition/conversation/cognitiveservices/v1?language=%s&format=detailed", s.cfg.AzureRegion, url.QueryEscape(locale))
|
||||
req, err := http.NewRequestWithContext(ctx, http.MethodPost, endpoint, bytes.NewReader(audio))
|
||||
if err != nil {
|
||||
return "", err
|
||||
}
|
||||
req.Header.Set("Ocp-Apim-Subscription-Key", s.cfg.AzureSpeechKey)
|
||||
req.Header.Set("Content-Type", contentTypeForAzure(contentType))
|
||||
req.Header.Set("Accept", "application/json")
|
||||
resp, err := s.client.Do(req)
|
||||
if err != nil {
|
||||
return "", err
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
if resp.StatusCode != http.StatusOK {
|
||||
return "", fmt.Errorf("Azure transcription HTTP %d", resp.StatusCode)
|
||||
}
|
||||
var result struct {
|
||||
Status string `json:"RecognitionStatus"`
|
||||
Text string `json:"DisplayText"`
|
||||
NBest []struct {
|
||||
Display string `json:"Display"`
|
||||
} `json:"NBest"`
|
||||
}
|
||||
if err = json.NewDecoder(io.LimitReader(resp.Body, 1<<20)).Decode(&result); err != nil {
|
||||
return "", err
|
||||
}
|
||||
if result.Status != "Success" {
|
||||
return "", errors.New("no speech recognized; please record again")
|
||||
}
|
||||
if len(result.NBest) > 0 && result.NBest[0].Display != "" {
|
||||
return result.NBest[0].Display, nil
|
||||
}
|
||||
if strings.TrimSpace(result.Text) == "" {
|
||||
return "", errors.New("no speech recognized")
|
||||
}
|
||||
return result.Text, nil
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user