From 7b11e31b4a4584c384cf92d3f5ada38c560eea9e Mon Sep 17 00:00:00 2001
From: Sun Yuhan <sunyuhan1998@users.noreply.github.com>
Date: Tue, 17 Jun 2025 10:06:03 +0800
Subject: [PATCH] fix: Fixed GH-3557, Fixed the issue where the audio filename
was lost when requesting the OpenAI `/transcriptions` and `/translations`
interfaces.
Signed-off-by: Sun Yuhan <sunyuhan1998@users.noreply.github.com>
---
.../openai/OpenAiAudioTranscriptionModel.java | 4 ++-
.../ai/openai/api/OpenAiAudioApi.java | 29 +++++++++++++++----
.../ai/openai/audio/api/OpenAiAudioApiIT.java | 19 +++++++-----
3 files changed, 39 insertions(+), 13 deletions(-)
diff --git a/models/spring-ai-openai/src/main/java/org/springframework/ai/openai/OpenAiAudioTranscriptionModel.java b/models/spring-ai-openai/src/main/java/org/springframework/ai/openai/OpenAiAudioTranscriptionModel.java
index 4c7bb105648..2773e163c0e 100644
--- a/models/spring-ai-openai/src/main/java/org/springframework/ai/openai/OpenAiAudioTranscriptionModel.java
+++ b/models/spring-ai-openai/src/main/java/org/springframework/ai/openai/OpenAiAudioTranscriptionModel.java
@@ -167,8 +167,10 @@ OpenAiAudioApi.TranscriptionRequest createRequest(AudioTranscriptionPrompt trans
}
}
+ Resource instructions = transcriptionPrompt.getInstructions();
return OpenAiAudioApi.TranscriptionRequest.builder()
- .file(toBytes(transcriptionPrompt.getInstructions()))
+ .file(toBytes(instructions))
+ .fileName(instructions.getFilename())
.responseFormat(options.getResponseFormat())
.prompt(options.getPrompt())
.temperature(options.getTemperature())
diff --git a/models/spring-ai-openai/src/main/java/org/springframework/ai/openai/api/OpenAiAudioApi.java b/models/spring-ai-openai/src/main/java/org/springframework/ai/openai/api/OpenAiAudioApi.java
index 1177a98f1d3..44e08cc114d 100644
--- a/models/spring-ai-openai/src/main/java/org/springframework/ai/openai/api/OpenAiAudioApi.java
+++ b/models/spring-ai-openai/src/main/java/org/springframework/ai/openai/api/OpenAiAudioApi.java
@@ -160,7 +160,7 @@ public <T> ResponseEntity<T> createTranscription(TranscriptionRequest requestBod
@Override
public String getFilename() {
- return "audio.webm";
+ return requestBody.fileName();
}
});
multipartBody.add("model", requestBody.model());
@@ -206,7 +206,7 @@ public <T> ResponseEntity<T> createTranslation(TranslationRequest requestBody, C
@Override
public String getFilename() {
- return "audio.webm";
+ return requestBody.fileName();
}
});
multipartBody.add("model", requestBody.model());
@@ -496,6 +496,7 @@ public SpeechRequest build() {
* Transcription</a>
*
* @param file The audio file to transcribe. Must be a valid audio file type.
+ * @param fileName The audio file name.
* @param model ID of the model to use. Only whisper-1 is currently available.
* @param language The language of the input audio. Supplying the input language in
* ISO-639-1 format will improve accuracy and latency.
@@ -517,6 +518,7 @@ public SpeechRequest build() {
public record TranscriptionRequest(
// @formatter:off
@JsonProperty("file") byte[] file,
+ @JsonProperty("fileName") String fileName,
@JsonProperty("model") String model,
@JsonProperty("language") String language,
@JsonProperty("prompt") String prompt,
@@ -554,6 +556,8 @@ public static class Builder {
private byte[] file;
+ private String fileName;
+
private String model = WhisperModel.WHISPER_1.getValue();
private String language;
@@ -571,6 +575,11 @@ public Builder file(byte[] file) {
return this;
}
+ public Builder fileName(String fileName) {
+ this.fileName = fileName;
+ return this;
+ }
+
public Builder model(String model) {
this.model = model;
return this;
@@ -603,11 +612,12 @@ public Builder granularityType(GranularityType granularityType) {
public TranscriptionRequest build() {
Assert.notNull(this.file, "file must not be null");
+ Assert.notNull(this.fileName, "fileName must not be null");
Assert.hasText(this.model, "model must not be empty");
Assert.notNull(this.responseFormat, "response_format must not be null");
- return new TranscriptionRequest(this.file, this.model, this.language, this.prompt, this.responseFormat,
- this.temperature, this.granularityType);
+ return new TranscriptionRequest(this.file, this.fileName, this.model, this.language, this.prompt,
+ this.responseFormat, this.temperature, this.granularityType);
}
}
@@ -619,6 +629,7 @@ public TranscriptionRequest build() {
*
* @param file The audio file object (not file name) to translate, in one of these
* formats: flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm.
+ * @param fileName The audio file name.
* @param model ID of the model to use. Only whisper-1 is currently available.
* @param prompt An optional text to guide the model's style or continue a previous
* audio segment. The prompt should be in English.
@@ -633,6 +644,7 @@ public TranscriptionRequest build() {
public record TranslationRequest(
// @formatter:off
@JsonProperty("file") byte[] file,
+ @JsonProperty("fileName") String fileName,
@JsonProperty("model") String model,
@JsonProperty("prompt") String prompt,
@JsonProperty("response_format") TranscriptResponseFormat responseFormat,
@@ -647,6 +659,8 @@ public static class Builder {
private byte[] file;
+ private String fileName;
+
private String model = WhisperModel.WHISPER_1.getValue();
private String prompt;
@@ -660,6 +674,11 @@ public Builder file(byte[] file) {
return this;
}
+ public Builder fileName(String fileName) {
+ this.fileName = fileName;
+ return this;
+ }
+
public Builder model(String model) {
this.model = model;
return this;
@@ -685,7 +704,7 @@ public TranslationRequest build() {
Assert.hasText(this.model, "model must not be empty");
Assert.notNull(this.responseFormat, "response_format must not be null");
- return new TranslationRequest(this.file, this.model, this.prompt, this.responseFormat,
+ return new TranslationRequest(this.file, this.fileName, this.model, this.prompt, this.responseFormat,
this.temperature);
}
diff --git a/models/spring-ai-openai/src/test/java/org/springframework/ai/openai/audio/api/OpenAiAudioApiIT.java b/models/spring-ai-openai/src/test/java/org/springframework/ai/openai/audio/api/OpenAiAudioApiIT.java
index 6c933dec283..6533d15de56 100644
--- a/models/spring-ai-openai/src/test/java/org/springframework/ai/openai/audio/api/OpenAiAudioApiIT.java
+++ b/models/spring-ai-openai/src/test/java/org/springframework/ai/openai/audio/api/OpenAiAudioApiIT.java
@@ -63,24 +63,29 @@ void speechTranscriptionAndTranslation() throws IOException {
FileCopyUtils.copy(speech, new File("target/speech.mp3"));
StructuredResponse translation = this.audioApi
- .createTranslation(
- TranslationRequest.builder().model(WhisperModel.WHISPER_1.getValue()).file(speech).build(),
- StructuredResponse.class)
+ .createTranslation(TranslationRequest.builder()
+ .model(WhisperModel.WHISPER_1.getValue())
+ .file(speech)
+ .fileName("speech.mp3")
+ .build(), StructuredResponse.class)
.getBody();
assertThat(translation.text().replaceAll(",", "")).isEqualTo("Hello my name is Chris and I love Spring AI.");
StructuredResponse transcriptionEnglish = this.audioApi
- .createTranscription(
- TranscriptionRequest.builder().model(WhisperModel.WHISPER_1.getValue()).file(speech).build(),
- StructuredResponse.class)
+ .createTranscription(TranscriptionRequest.builder()
+ .model(WhisperModel.WHISPER_1.getValue())
+ .file(speech)
+ .fileName("speech.mp3")
+ .build(), StructuredResponse.class)
.getBody();
assertThat(transcriptionEnglish.text().replaceAll(",", ""))
.isEqualTo("Hello my name is Chris and I love Spring AI.");
StructuredResponse transcriptionDutch = this.audioApi
- .createTranscription(TranscriptionRequest.builder().file(speech).language("nl").build(),
+ .createTranscription(
+ TranscriptionRequest.builder().file(speech).fileName("speech.mp3").language("nl").build(),
StructuredResponse.class)
.getBody();