A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 8, 20:15 UTC
[Misc] Rename `translations` to `speech_to_text` for OAI serving component (#33904)
Signed-off-by: NickLucche <nlucches@redhat.com>
Nicolò Lucchesi committed
Feb 5, 2026 at 20:16 UTC
20f5d185a6f570b74ab403577cd4eabe44d14496
7 files changed
+11
-11
vllm/entrypoints/openai/api_server.py
+4
-4
@@ -190,11 +190,11 @@ def build_app(args: Namespace, supported_tasks: tuple["SupportedTask", ...]) ->
190
register_generate_api_routers(app)
191
192
if "transcription" in supported_tasks:
193
- from vllm.entrypoints.openai.translations.api_router import (
194
- attach_router as register_translations_api_router,
193
+ from vllm.entrypoints.openai.speech_to_text.api_router import (
194
+ attach_router as register_speech_to_text_api_router,
195
)
196
197
- register_translations_api_router(app)
197
+ register_speech_to_text_api_router(app)
198
199
if "realtime" in supported_tasks:
200
from vllm.entrypoints.openai.realtime.api_router import (
@@ -318,7 +318,7 @@ async def init_app_state(
318
)
319
320
if "transcription" in supported_tasks:
321
- from vllm.entrypoints.openai.translations.api_router import (
321
+ from vllm.entrypoints.openai.speech_to_text.api_router import (
322
init_transcription_state,
323
)
324
vllm/entrypoints/openai/engine/serving.py
+1
-1
@@ -57,7 +57,7 @@ from vllm.entrypoints.openai.responses.protocol import (
57
from vllm.entrypoints.openai.responses.utils import (
58
construct_input_messages,
59
)
60
-from vllm.entrypoints.openai.translations.protocol import (
60
+from vllm.entrypoints.openai.speech_to_text.protocol import (
61
TranscriptionRequest,
62
TranscriptionResponse,
63
TranslationRequest,
vllm/entrypoints/openai/speech_to_text/__init__.py
renamed
vllm/entrypoints/openai/speech_to_text/api_router.py
renamed
+2
-2
@@ -9,13 +9,13 @@ from fastapi import APIRouter, FastAPI, Form, Request
9
from fastapi.responses import JSONResponse, StreamingResponse
10
11
from vllm.entrypoints.openai.engine.protocol import ErrorResponse
12
-from vllm.entrypoints.openai.translations.protocol import (
12
+from vllm.entrypoints.openai.speech_to_text.protocol import (
13
TranscriptionRequest,
14
TranscriptionResponseVariant,
15
TranslationRequest,
16
TranslationResponseVariant,
17
)
18
-from vllm.entrypoints.openai.translations.serving import (
18
+from vllm.entrypoints.openai.speech_to_text.serving import (
19
OpenAIServingTranscription,
20
OpenAIServingTranslation,
21
)
vllm/entrypoints/openai/speech_to_text/protocol.py
renamed
vllm/entrypoints/openai/speech_to_text/serving.py
renamed
+2
-2
@@ -11,7 +11,7 @@ from vllm.entrypoints.openai.engine.protocol import (
11
RequestResponseMetadata,
12
)
13
from vllm.entrypoints.openai.models.serving import OpenAIServingModels
14
-from vllm.entrypoints.openai.translations.protocol import (
14
+from vllm.entrypoints.openai.speech_to_text.protocol import (
15
TranscriptionRequest,
16
TranscriptionResponse,
17
TranscriptionResponseStreamChoice,
@@ -23,7 +23,7 @@ from vllm.entrypoints.openai.translations.protocol import (
23
TranslationResponseVerbose,
24
TranslationStreamResponse,
25
)
26
-from vllm.entrypoints.openai.translations.speech_to_text import OpenAISpeechToText
26
+from vllm.entrypoints.openai.speech_to_text.speech_to_text import OpenAISpeechToText
27
from vllm.logger import init_logger
28
from vllm.outputs import RequestOutput
29
vllm/entrypoints/openai/speech_to_text/speech_to_text.py
renamed
+2
-2
@@ -24,7 +24,7 @@ from vllm.entrypoints.openai.engine.protocol import (
24
)
25
from vllm.entrypoints.openai.engine.serving import OpenAIServing, SpeechToTextRequest
26
from vllm.entrypoints.openai.models.serving import OpenAIServingModels
27
-from vllm.entrypoints.openai.translations.protocol import (
27
+from vllm.entrypoints.openai.speech_to_text.protocol import (
28
TranscriptionResponse,
29
TranscriptionResponseStreamChoice,
30
TranscriptionResponseVerbose,
@@ -402,7 +402,7 @@ class OpenAISpeechToText(OpenAIServing):
402
audio_data: bytes,
403
request: SpeechToTextRequest,
404
raw_request: Request,
405
- response_class: type[T | V],
405
+ response_class: type[ResponseType],
406
stream_generator_method: Callable[..., AsyncGenerator[str, None]],
407
) -> T | V | AsyncGenerator[str, None] | ErrorResponse:
408
"""Base method for speech-to-text operations like transcription and