diff --git a/src/main/java/com/debatetimer/service/transcript/SpeechTranscribeService.java b/src/main/java/com/debatetimer/service/transcript/SpeechTranscribeService.java
index 6a7218b0..93a454d7 100644
--- a/src/main/java/com/debatetimer/service/transcript/SpeechTranscribeService.java
+++ b/src/main/java/com/debatetimer/service/transcript/SpeechTranscribeService.java
@@ -5,9 +5,11 @@
 import com.debatetimer.domain.transcript.UtteranceAssembler;
 import com.debatetimer.dto.transcript.response.SpeechTranscribeResponse;
 import com.debatetimer.exception.custom.DTClientErrorException;
+import com.debatetimer.exception.custom.DTServerErrorException;
 import com.debatetimer.exception.errorcode.ClientErrorCode;
 import java.util.List;
 import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
 import org.springframework.lang.Nullable;
 import org.springframework.stereotype.Service;
 import org.springframework.web.multipart.MultipartFile;
@@ -16,6 +18,7 @@
  * 토론 기록 Soniox 모드에서 타이머가 끝난 발언 구간의 오디오를 받아 문장 단위로 전사한다.
  * Soniox 전사가 끝날 때까지 요청 스레드에서 기다린다.
  */
+@Slf4j
 @Service
 @RequiredArgsConstructor
 public class SpeechTranscribeService {
@@ -27,10 +30,21 @@ public SpeechTranscribeResponse transcribeWithSoniox(MultipartFile audio, @Nulla
         if (audio == null || audio.isEmpty()) {
             throw new DTClientErrorException(ClientErrorCode.EMPTY_AUDIO);
         }
-        List<SpeechToken> tokens = sonioxClient.transcribe(audio.getResource(), toLanguageHints(language));
+        List<SpeechToken> tokens = transcribe(audio, language);
         return new SpeechTranscribeResponse(utteranceAssembler.assemble(tokens));
     }
 
+    // Soniox가 오디오를 거절하면 어떤 파일이었는지 알 수 있도록 업로드한 오디오 정보를 함께 남긴다.
+    private List<SpeechToken> transcribe(MultipartFile audio, @Nullable String language) {
+        try {
+            return sonioxClient.transcribe(audio.getResource(), toLanguageHints(language));
+        } catch (DTServerErrorException exception) {
+            log.warn("Soniox 전사에 실패한 오디오입니다. name={}, contentType={}, size={}",
+                    audio.getOriginalFilename(), audio.getContentType(), audio.getSize());
+            throw exception;
+        }
+    }
+
     // 프론트 음성 인식 언어(ko-KR, en-US 등)를 Soniox 언어 힌트로 바꾼다. 기본은 한국어다.
     private List<String> toLanguageHints(@Nullable String language) {
         if (language != null && language.toLowerCase().startsWith("en")) {
diff --git a/src/test/java/com/debatetimer/service/transcript/SpeechTranscribeServiceTest.java b/src/test/java/com/debatetimer/service/transcript/SpeechTranscribeServiceTest.java
index 9f0296f5..10a536ce 100644
--- a/src/test/java/com/debatetimer/service/transcript/SpeechTranscribeServiceTest.java
+++ b/src/test/java/com/debatetimer/service/transcript/SpeechTranscribeServiceTest.java
@@ -13,15 +13,21 @@
 import com.debatetimer.domain.transcript.Utterance;
 import com.debatetimer.dto.transcript.response.SpeechTranscribeResponse;
 import com.debatetimer.exception.custom.DTClientErrorException;
+import com.debatetimer.exception.custom.DTServerErrorException;
+import com.debatetimer.exception.errorcode.ServerErrorCode;
 import com.debatetimer.exception.errorcode.ClientErrorCode;
 import com.debatetimer.service.BaseServiceTest;
 import java.util.List;
 import org.junit.jupiter.api.Nested;
 import org.junit.jupiter.api.Test;
+import org.junit.jupiter.api.extension.ExtendWith;
+import org.springframework.boot.test.system.CapturedOutput;
+import org.springframework.boot.test.system.OutputCaptureExtension;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.mock.web.MockMultipartFile;
 import org.springframework.test.context.bean.override.mockito.MockitoBean;
 
+@ExtendWith(OutputCaptureExtension.class)
 class SpeechTranscribeServiceTest extends BaseServiceTest {
 
     private static final MockMultipartFile AUDIO = new MockMultipartFile("audio", "segment.webm", "audio/webm",
@@ -70,5 +76,19 @@ class TranscribeWithSoniox {
                     .hasMessage(ClientErrorCode.EMPTY_AUDIO.getMessage());
             verifyNoInteractions(sonioxClient);
         }
+
+        @Test
+        void 전사에_실패하면_오디오_이름_형식_크기를_로그에_남기고_예외를_그대로_던진다(CapturedOutput output) {
+            when(sonioxClient.transcribe(any(), any()))
+                    .thenThrow(new DTServerErrorException(ServerErrorCode.STT_REQUEST_FAILED));
+
+            assertThatThrownBy(() -> speechTranscribeService.transcribeWithSoniox(AUDIO, "ko-KR"))
+                    .isInstanceOf(DTServerErrorException.class)
+                    .hasMessage(ServerErrorCode.STT_REQUEST_FAILED.getMessage());
+            assertThat(output.getAll())
+                    .contains("segment.webm")
+                    .contains("audio/webm")
+                    .contains("size=3");
+        }
     }
 }