add openai create speech endpoint (#305)

Signed-off-by: Bhanu068 <voutharoja.bhanu06@gmail.com>
Signed-off-by: Bhanu Prakash Voutharoja <59905694+Bhanu068@users.noreply.github.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>
Co-authored-by: Yueqian Lin <70319226+linyueqian@users.noreply.github.com>
This commit is contained in:
authored and GitHub committed 2025-12-27 13:56:51 +08:00
1 parent fb57a6162f
commit 2f69748ee4
6 files changed
+527 -16

No files matched your search

@@ -46,12 +46,14 @@ from vllm_omni.entrypoints.openai.image_api_utils import (
encode_image_base64,
parse_size,
)
from vllm_omni.entrypoints.openai.protocol.audio import OpenAICreateSpeechRequest
from vllm_omni.entrypoints.openai.protocol.images import (
ImageData,
ImageGenerationRequest,
ImageGenerationResponse,
)
from vllm_omni.entrypoints.openai.serving_chat import OmniOpenAIServingChat
from vllm_omni.entrypoints.openai.serving_speech import OmniOpenAIServingSpeech
logger = init_logger(__name__)
@@ -432,6 +434,10 @@ async def omni_init_app_state(
log_error_stack=args.log_error_stack,
)
state.openai_serving_speech = OmniOpenAIServingSpeech(
engine_client, state.openai_serving_models, request_logger=request_logger
)
state.enable_server_load_tracking = args.enable_server_load_tracking
state.server_load_metrics = 0
@@ -440,6 +446,10 @@ def Omnichat(request: Request) -> OmniOpenAIServingChat | None:
return request.app.state.openai_serving_chat
def Omnispeech(request: Request) -> OmniOpenAIServingSpeech | None:
return request.app.state.openai_serving_speech
@router.post(
"/v1/chat/completions",
dependencies=[Depends(validate_json_request)],
@@ -496,6 +506,28 @@ async def create_chat_completion(request: ChatCompletionRequest, raw_request: Re
return StreamingResponse(content=generator, media_type="text/event-stream")
@router.post(
"/v1/audio/speech",
dependencies=[Depends(validate_json_request)],
responses={
HTTPStatus.OK.value: {"content": {"audio/*": {}}},
HTTPStatus.BAD_REQUEST.value: {"model": ErrorResponse},
HTTPStatus.NOT_FOUND.value: {"model": ErrorResponse},
HTTPStatus.INTERNAL_SERVER_ERROR.value: {"model": ErrorResponse},
},
)
@with_cancellation
@load_aware_call
async def create_speech(request: OpenAICreateSpeechRequest, raw_request: Request):
handler = Omnispeech(raw_request)
if handler is None:
return base(raw_request).create_error_response(message="The model does not support Speech API")
try:
return await handler.create_speech(request, raw_request)
except Exception as e:
raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value, detail=str(e)) from e
# Image generation API endpoints