[ASR] Add Long Audio benchmark and correctness test (#44587)

Signed-off-by: Ekagra Ranjan <3116519+ekagra-ranjan@users.noreply.github.com>
This commit is contained in:
Ekagra Ranjan
2026-06-12 04:11:16 +00:00
committed by GitHub
parent b927004c44
commit 226ba9fc9e
5 changed files with 530 additions and 46 deletions
+44 -11
View File
@@ -445,7 +445,6 @@ async def async_request_openai_audio(
api_url = request_func_input.api_url
_validate_api_url(api_url, "OpenAI Audio API", {"transcriptions", "translations"})
content = [{"type": "text", "text": request_func_input.prompt}]
payload = {
"model": request_func_input.model_name
if request_func_input.model_name
@@ -469,19 +468,26 @@ async def async_request_openai_audio(
buffer.seek(0)
return buffer
mm_audio = request_func_input.multi_modal_content
if not isinstance(mm_audio, dict) or "audio" not in mm_audio:
raise TypeError("multi_modal_content must be a dict containing 'audio'")
with to_bytes(*mm_audio["audio"]) as f:
async def send_audio_file(
audio_file: io.BytesIO | Any,
*,
input_audio_duration: float,
filename: str | None = None,
content_type: str | None = None,
) -> RequestFuncOutput:
form = aiohttp.FormData()
form.add_field("file", f, content_type="audio/wav")
add_field_kwargs: dict[str, str] = {}
if filename is not None:
add_field_kwargs["filename"] = filename
if content_type is not None:
add_field_kwargs["content_type"] = content_type
form.add_field("file", audio_file, **add_field_kwargs)
for key, value in payload.items():
form.add_field(key, str(value))
output = RequestFuncOutput()
output.prompt_len = request_func_input.prompt_len
output.input_audio_duration = soundfile.info(f).duration
f.seek(0)
output.input_audio_duration = input_audio_duration
generated_text = ""
ttft = 0.0
@@ -541,9 +547,36 @@ async def async_request_openai_audio(
exc_info = sys.exc_info()
output.error = "".join(traceback.format_exception(*exc_info))
if pbar:
pbar.update(1)
return output
if pbar:
pbar.update(1)
return output
mm_audio = request_func_input.multi_modal_content
if not isinstance(mm_audio, dict):
raise TypeError(
"multi_modal_content must be a dict containing 'audio' or 'audio_path'"
)
if "audio" in mm_audio:
with to_bytes(*mm_audio["audio"]) as f:
input_audio_duration = soundfile.info(f).duration
f.seek(0)
return await send_audio_file(
f,
input_audio_duration=input_audio_duration,
filename="audio.wav",
content_type="audio/wav",
)
if "audio_path" in mm_audio:
audio_path = mm_audio["audio_path"]
with open(audio_path, "rb") as f:
return await send_audio_file(
f,
input_audio_duration=soundfile.info(audio_path).duration,
filename=os.path.basename(audio_path),
)
raise TypeError(
"multi_modal_content must be a dict containing 'audio' or 'audio_path'"
)
async def _run_pooling_request(