Spaces:
Configuration error
Configuration error
Fedir Zadniprovskyi
commited on
Commit
·
ed63b7d
1
Parent(s):
b8804a6
fix: `timestamp_granularities[]` handling (#28, #58, #81)
Browse files
src/faster_whisper_server/routers/stt.py
CHANGED
@@ -9,6 +9,7 @@ from fastapi import (
|
|
9 |
APIRouter,
|
10 |
Form,
|
11 |
Query,
|
|
|
12 |
Response,
|
13 |
UploadFile,
|
14 |
WebSocket,
|
@@ -30,6 +31,8 @@ from faster_whisper_server.config import (
|
|
30 |
from faster_whisper_server.core import Segment, segments_to_srt, segments_to_text, segments_to_vtt
|
31 |
from faster_whisper_server.dependencies import ConfigDependency, ModelManagerDependency, get_config
|
32 |
from faster_whisper_server.server_models import (
|
|
|
|
|
33 |
TimestampGranularities,
|
34 |
TranscriptionJsonResponse,
|
35 |
TranscriptionVerboseJsonResponse,
|
@@ -150,6 +153,18 @@ def translate_file(
|
|
150 |
return segments_to_response(segments, transcription_info, response_format)
|
151 |
|
152 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
153 |
# https://platform.openai.com/docs/api-reference/audio/createTranscription
|
154 |
# https://github.com/openai/openai-openapi/blob/master/openapi.yaml#L8915
|
155 |
@router.post(
|
@@ -159,6 +174,7 @@ def translate_file(
|
|
159 |
def transcribe_file(
|
160 |
config: ConfigDependency,
|
161 |
model_manager: ModelManagerDependency,
|
|
|
162 |
file: Annotated[UploadFile, Form()],
|
163 |
model: Annotated[ModelName | None, Form()] = None,
|
164 |
language: Annotated[Language | None, Form()] = None,
|
@@ -167,6 +183,7 @@ def transcribe_file(
|
|
167 |
temperature: Annotated[float, Form()] = 0.0,
|
168 |
timestamp_granularities: Annotated[
|
169 |
TimestampGranularities,
|
|
|
170 |
Form(alias="timestamp_granularities[]"),
|
171 |
] = ["segment"],
|
172 |
stream: Annotated[bool, Form()] = False,
|
@@ -178,6 +195,11 @@ def transcribe_file(
|
|
178 |
language = config.default_language
|
179 |
if response_format is None:
|
180 |
response_format = config.default_response_format
|
|
|
|
|
|
|
|
|
|
|
181 |
whisper = model_manager.load_model(model)
|
182 |
segments, transcription_info = whisper.transcribe(
|
183 |
file.file,
|
|
|
9 |
APIRouter,
|
10 |
Form,
|
11 |
Query,
|
12 |
+
Request,
|
13 |
Response,
|
14 |
UploadFile,
|
15 |
WebSocket,
|
|
|
31 |
from faster_whisper_server.core import Segment, segments_to_srt, segments_to_text, segments_to_vtt
|
32 |
from faster_whisper_server.dependencies import ConfigDependency, ModelManagerDependency, get_config
|
33 |
from faster_whisper_server.server_models import (
|
34 |
+
DEFAULT_TIMESTAMP_GRANULARITIES,
|
35 |
+
TIMESTAMP_GRANULARITIES_COMBINATIONS,
|
36 |
TimestampGranularities,
|
37 |
TranscriptionJsonResponse,
|
38 |
TranscriptionVerboseJsonResponse,
|
|
|
153 |
return segments_to_response(segments, transcription_info, response_format)
|
154 |
|
155 |
|
156 |
+
# HACK: Since Form() doesn't support `alias`, we need to use a workaround.
|
157 |
+
async def get_timestamp_granularities(request: Request) -> TimestampGranularities:
|
158 |
+
form = await request.form()
|
159 |
+
if form.get("timestamp_granularities[]") is None:
|
160 |
+
return DEFAULT_TIMESTAMP_GRANULARITIES
|
161 |
+
timestamp_granularities = form.getlist("timestamp_granularities[]")
|
162 |
+
assert (
|
163 |
+
timestamp_granularities in TIMESTAMP_GRANULARITIES_COMBINATIONS
|
164 |
+
), f"{timestamp_granularities} is not a valid value for `timestamp_granularities[]`."
|
165 |
+
return timestamp_granularities
|
166 |
+
|
167 |
+
|
168 |
# https://platform.openai.com/docs/api-reference/audio/createTranscription
|
169 |
# https://github.com/openai/openai-openapi/blob/master/openapi.yaml#L8915
|
170 |
@router.post(
|
|
|
174 |
def transcribe_file(
|
175 |
config: ConfigDependency,
|
176 |
model_manager: ModelManagerDependency,
|
177 |
+
request: Request,
|
178 |
file: Annotated[UploadFile, Form()],
|
179 |
model: Annotated[ModelName | None, Form()] = None,
|
180 |
language: Annotated[Language | None, Form()] = None,
|
|
|
183 |
temperature: Annotated[float, Form()] = 0.0,
|
184 |
timestamp_granularities: Annotated[
|
185 |
TimestampGranularities,
|
186 |
+
# WARN: `alias` doesn't actually work.
|
187 |
Form(alias="timestamp_granularities[]"),
|
188 |
] = ["segment"],
|
189 |
stream: Annotated[bool, Form()] = False,
|
|
|
195 |
language = config.default_language
|
196 |
if response_format is None:
|
197 |
response_format = config.default_response_format
|
198 |
+
timestamp_granularities = asyncio.run(get_timestamp_granularities(request))
|
199 |
+
if timestamp_granularities != DEFAULT_TIMESTAMP_GRANULARITIES and response_format != ResponseFormat.VERBOSE_JSON:
|
200 |
+
logger.warning(
|
201 |
+
"It only makes sense to provide `timestamp_granularities[]` when `response_format` is set to `verbose_json`. See https://platform.openai.com/docs/api-reference/audio/createTranscription#audio-createtranscription-timestamp_granularities." # noqa: E501
|
202 |
+
)
|
203 |
whisper = model_manager.load_model(model)
|
204 |
segments, transcription_info = whisper.transcribe(
|
205 |
file.file,
|
src/faster_whisper_server/server_models.py
CHANGED
@@ -29,7 +29,7 @@ class TranscriptionVerboseJsonResponse(BaseModel):
|
|
29 |
language: str
|
30 |
duration: float
|
31 |
text: str
|
32 |
-
words: list[Word]
|
33 |
segments: list[Segment]
|
34 |
|
35 |
@classmethod
|
@@ -38,7 +38,7 @@ class TranscriptionVerboseJsonResponse(BaseModel):
|
|
38 |
language=transcription_info.language,
|
39 |
duration=segment.end - segment.start,
|
40 |
text=segment.text,
|
41 |
-
words=
|
42 |
segments=[segment],
|
43 |
)
|
44 |
|
@@ -51,7 +51,7 @@ class TranscriptionVerboseJsonResponse(BaseModel):
|
|
51 |
duration=transcription_info.duration,
|
52 |
text=segments_to_text(segments),
|
53 |
segments=segments,
|
54 |
-
words=Word.from_segments(segments),
|
55 |
)
|
56 |
|
57 |
@classmethod
|
@@ -112,6 +112,7 @@ class ModelObject(BaseModel):
|
|
112 |
TimestampGranularities = list[Literal["segment", "word"]]
|
113 |
|
114 |
|
|
|
115 |
TIMESTAMP_GRANULARITIES_COMBINATIONS: list[TimestampGranularities] = [
|
116 |
[], # should be treated as ["segment"]. https://platform.openai.com/docs/api-reference/audio/createTranscription#audio-createtranscription-timestamp_granularities
|
117 |
["segment"],
|
|
|
29 |
language: str
|
30 |
duration: float
|
31 |
text: str
|
32 |
+
words: list[Word] | None
|
33 |
segments: list[Segment]
|
34 |
|
35 |
@classmethod
|
|
|
38 |
language=transcription_info.language,
|
39 |
duration=segment.end - segment.start,
|
40 |
text=segment.text,
|
41 |
+
words=segment.words if transcription_info.transcription_options.word_timestamps else None,
|
42 |
segments=[segment],
|
43 |
)
|
44 |
|
|
|
51 |
duration=transcription_info.duration,
|
52 |
text=segments_to_text(segments),
|
53 |
segments=segments,
|
54 |
+
words=Word.from_segments(segments) if transcription_info.transcription_options.word_timestamps else None,
|
55 |
)
|
56 |
|
57 |
@classmethod
|
|
|
112 |
TimestampGranularities = list[Literal["segment", "word"]]
|
113 |
|
114 |
|
115 |
+
DEFAULT_TIMESTAMP_GRANULARITIES: TimestampGranularities = ["segment"]
|
116 |
TIMESTAMP_GRANULARITIES_COMBINATIONS: list[TimestampGranularities] = [
|
117 |
[], # should be treated as ["segment"]. https://platform.openai.com/docs/api-reference/audio/createTranscription#audio-createtranscription-timestamp_granularities
|
118 |
["segment"],
|
tests/api_timestamp_granularities_test.py
ADDED
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
"""See `tests/openai_timestamp_granularities_test.py` to understand how OpenAI handles `response_type` and `timestamp_granularities`.""" # noqa: E501
|
2 |
+
|
3 |
+
from faster_whisper_server.server_models import TIMESTAMP_GRANULARITIES_COMBINATIONS, TimestampGranularities
|
4 |
+
from openai import AsyncOpenAI
|
5 |
+
import pytest
|
6 |
+
|
7 |
+
|
8 |
+
@pytest.mark.asyncio()
|
9 |
+
@pytest.mark.parametrize("timestamp_granularities", TIMESTAMP_GRANULARITIES_COMBINATIONS)
|
10 |
+
async def test_api_json_response_format_and_timestamp_granularities_combinations(
|
11 |
+
openai_client: AsyncOpenAI,
|
12 |
+
timestamp_granularities: TimestampGranularities,
|
13 |
+
) -> None:
|
14 |
+
audio_file = open("audio.wav", "rb") # noqa: SIM115, ASYNC230
|
15 |
+
|
16 |
+
await openai_client.audio.transcriptions.create(
|
17 |
+
file=audio_file, model="whisper-1", response_format="json", timestamp_granularities=timestamp_granularities
|
18 |
+
)
|
19 |
+
|
20 |
+
|
21 |
+
@pytest.mark.asyncio()
|
22 |
+
@pytest.mark.parametrize("timestamp_granularities", TIMESTAMP_GRANULARITIES_COMBINATIONS)
|
23 |
+
async def test_api_verbose_json_response_format_and_timestamp_granularities_combinations(
|
24 |
+
openai_client: AsyncOpenAI,
|
25 |
+
timestamp_granularities: TimestampGranularities,
|
26 |
+
) -> None:
|
27 |
+
audio_file = open("audio.wav", "rb") # noqa: SIM115, ASYNC230
|
28 |
+
|
29 |
+
transcription = await openai_client.audio.transcriptions.create(
|
30 |
+
file=audio_file,
|
31 |
+
model="whisper-1",
|
32 |
+
response_format="verbose_json",
|
33 |
+
timestamp_granularities=timestamp_granularities,
|
34 |
+
)
|
35 |
+
|
36 |
+
assert transcription.__pydantic_extra__
|
37 |
+
if "word" in timestamp_granularities:
|
38 |
+
assert transcription.__pydantic_extra__.get("segments") is not None
|
39 |
+
assert transcription.__pydantic_extra__.get("words") is not None
|
40 |
+
else:
|
41 |
+
# Unless explicitly requested, words are not present
|
42 |
+
assert transcription.__pydantic_extra__.get("segments") is not None
|
43 |
+
assert transcription.__pydantic_extra__.get("words") is None
|