Segment Voxtral transcripts for telephony
All checks were successful
CI / test (push) Successful in 15s
Build and Deploy / build-and-deploy (push) Successful in 32s

This commit is contained in:
Grendgi
2026-06-09 16:12:57 +03:00
parent e6c2b46cf6
commit 64bf40b3ba
2 changed files with 180 additions and 12 deletions

View File

@@ -75,17 +75,24 @@ func TestVoxtralUsesAudioTranscriptionsEndpoint(t *testing.T) {
}))
defer audioSrv.Close()
var gotPath, gotModel string
var gotPath, gotModel, gotResponseFormat string
providerSrv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
gotPath = r.URL.Path
if err := r.ParseMultipartForm(16 << 20); err != nil {
t.Fatalf("ParseMultipartForm: %v", err)
}
gotModel = r.FormValue("model")
gotResponseFormat = r.FormValue("response_format")
if _, _, err := r.FormFile("file"); err != nil {
t.Fatalf("FormFile: %v", err)
}
_ = json.NewEncoder(w).Encode(map[string]string{"text": "Алло, тест."})
_ = json.NewEncoder(w).Encode(map[string]any{
"text": "Алло, тест. Да, слышно.",
"segments": []map[string]any{
{"start": 0, "end": 1.2, "text": "Алло, тест."},
{"start": 1.2, "end": 2.4, "text": "Да, слышно."},
},
})
}))
defer providerSrv.Close()
@@ -107,11 +114,27 @@ func TestVoxtralUsesAudioTranscriptionsEndpoint(t *testing.T) {
if gotModel != "mistralai/Voxtral-Small-24B-2507" {
t.Fatalf("model = %q", gotModel)
}
if len(got.Segments) != 1 || got.Segments[0].Text != "Алло, тест." {
if gotResponseFormat != "verbose_json" {
t.Fatalf("response_format = %q, want verbose_json", gotResponseFormat)
}
if len(got.Segments) != 2 || got.Segments[0].Text != "Алло, тест." || got.Segments[1].Start != 1.2 {
t.Fatalf("segments = %#v", got.Segments)
}
}
func TestSegmentTranscriptTextAddsHeuristicSpeakers(t *testing.T) {
got := segmentTranscriptText("Алло, добрый день. Да, слушаю. Скажите, квартира продается? Да, продается.", true)
if len(got) < 2 {
t.Fatalf("segments = %#v, want multiple", got)
}
if got[0].Speaker != "SPEAKER_00" || got[1].Speaker != "SPEAKER_01" {
t.Fatalf("speakers = %q/%q", got[0].Speaker, got[1].Speaker)
}
if got[1].Start <= got[0].Start {
t.Fatalf("segment times did not advance: %#v", got)
}
}
func near(got, want float64) bool {
return math.Abs(got-want) < 0.000001
}