Use Voxtral audio transcription endpoint

internal/transcription/client_test.go

@@ -1,7 +1,10 @@
 package transcription
 
 import (
+	"encoding/json"
 	"math"
+	"net/http"
+	"net/http/httptest"
 	"testing"
 	"time"
 )
@@ -66,6 +69,49 @@ func TestAudioDataURLUsesVLLMAudioURLFormat(t *testing.T) {
 	}
 }
 
+func TestVoxtralUsesAudioTranscriptionsEndpoint(t *testing.T) {
+	audioSrv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		_, _ = w.Write([]byte("fake audio"))
+	}))
+	defer audioSrv.Close()
+
+	var gotPath, gotModel 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")
+		if _, _, err := r.FormFile("file"); err != nil {
+			t.Fatalf("FormFile: %v", err)
+		}
+		_ = json.NewEncoder(w).Encode(map[string]string{"text": "Алло, тест."})
+	}))
+	defer providerSrv.Close()
+
+	client := NewWithOptions(Options{
+		Providers:      []string{"voxtral-small"},
+		VoxtralBaseURL: providerSrv.URL,
+		VoxtralModel:   "mistralai/Voxtral-Small-24B-2507",
+	})
+	if client == nil {
+		t.Fatal("client is nil")
+	}
+	got, err := client.Transcribe(t.Context(), Input{AudioURL: audioSrv.URL, Filename: "call.mp3"})
+	if err != nil {
+		t.Fatalf("Transcribe: %v", err)
+	}
+	if gotPath != "/v1/audio/transcriptions" {
+		t.Fatalf("path = %q, want /v1/audio/transcriptions", gotPath)
+	}
+	if gotModel != "mistralai/Voxtral-Small-24B-2507" {
+		t.Fatalf("model = %q", gotModel)
+	}
+	if len(got.Segments) != 1 || got.Segments[0].Text != "Алло, тест." {
+		t.Fatalf("segments = %#v", got.Segments)
+	}
+}
+
 func near(got, want float64) bool {
 	return math.Abs(got-want) < 0.000001
 }