Commit @39ee11644eedfd9dac27d3cf4d795597505e3622 - yjyoon/whisper_server

d0feed8

39ee116

.github/workflows/docker-build-and-push.yaml

--- .github/workflows/docker-build-and-push.yaml

+++ .github/workflows/docker-build-and-push.yaml


         uses: docker/metadata-action@v5
         with:
           images: |
-            fedirz/speaches
+            fedirz/faster-whisper-server
           # https://github.com/docker/metadata-action?tab=readme-ov-file#flavor-input
           flavor: |
             latest=false

           # platforms: linux/amd64,linux/arm64
           tags: ${{ steps.meta.outputs.tags }}
           # TODO: cache
-          # cache-from: type=registry,ref=fedirz/speaches:buildcache
-          # cache-to: type=registry,ref=fedirz/speaches:buildcache,mode=max
+          # cache-from: type=registry,ref=fedirz/faster-whisper-server:buildcache
+          # cache-to: type=registry,ref=fedirz/faster-whisper-server:buildcache,mode=max

d0feed8

39ee116

Dockerfile.cpu

--- Dockerfile.cpu

+++ Dockerfile.cpu


     rm -rf /var/lib/apt/lists/* && \
     curl -sS https://bootstrap.pypa.io/get-pip.py | python3.11
 RUN pip install --no-cache-dir poetry==1.8.2
-WORKDIR /root/speaches
+WORKDIR /root/faster-whisper-server
 COPY pyproject.toml poetry.lock ./
 RUN poetry install --only main
-COPY ./speaches ./speaches
+COPY ./faster_whisper_server ./faster_whisper_server
 ENTRYPOINT ["poetry", "run"]
-CMD ["uvicorn", "speaches.main:app"]
+CMD ["uvicorn", "faster_whisper_server.main:app"]
 ENV WHISPER_MODEL=distil-medium.en
 ENV WHISPER_INFERENCE_DEVICE=cpu
 ENV WHISPER_COMPUTE_TYPE=int8

d0feed8

39ee116

Dockerfile.cuda

--- Dockerfile.cuda

+++ Dockerfile.cuda


     rm -rf /var/lib/apt/lists/* && \
     curl -sS https://bootstrap.pypa.io/get-pip.py | python3.11
 RUN pip install --no-cache-dir poetry==1.8.2
-WORKDIR /root/speaches
+WORKDIR /root/faster-whisper-server
 COPY pyproject.toml poetry.lock ./
 RUN poetry install --only main
-COPY ./speaches ./speaches
+COPY ./faster_whisper_server ./faster_whisper_server
 ENTRYPOINT ["poetry", "run"]
-CMD ["uvicorn", "speaches.main:app"]
+CMD ["uvicorn", "faster_whisper_server.main:app"]
 ENV WHISPER_MODEL=distil-large-v3
 ENV WHISPER_INFERENCE_DEVICE=cuda
 ENV UVICORN_HOST=0.0.0.0

d0feed8

39ee116

README.md

--- README.md

+++ README.md


-# WARN: WIP (code is ugly, bad documentation, may have bugs, test files aren't included, CPU inference was barely tested, etc.)
-# Intro
-:peach:`speaches` is a web server that supports real-time transcription using WebSockets.
+## Faster Whisper Server
+`faster-whisper-server` is a web server that supports real-time transcription using WebSockets.
 - [faster-whisper](https://github.com/SYSTRAN/faster-whisper) is used as the backend. Both GPU and CPU inference are supported.
 - LocalAgreement2 ([paper](https://aclanthology.org/2023.ijcnlp-demo.3.pdf) | [original implementation](https://github.com/ufal/whisper_streaming)) algorithm is used for real-time transcription.
 - Can be deployed using Docker (Compose configuration can be found in [compose.yaml](./compose.yaml)).
-- All configuration is done through environment variables. See [config.py](./speaches/config.py).
+- All configuration is done through environment variables. See [config.py](./faster_whisper_server/config.py).
 - NOTE: only transcription of single channel, 16000 sample rate, raw, 16-bit little-endian audio is supported.
 - NOTE: this isn't really meant to be used as a standalone tool but rather to add transcription features to other applications.
 Please create an issue if you find a bug, have a question, or a feature suggestion.
 # Quick Start
-Spinning up a `speaches` web server
+Using Docker
 ```bash
-docker run --gpus=all --publish 8000:8000 --mount type=bind,source=$HOME/.cache/huggingface,target=/root/.cache/huggingface fedirz/speaches:cuda
+docker run --gpus=all --publish 8000:8000 --volume ~/.cache/huggingface:/root/.cache/huggingface fedirz/faster-whisper-server:cuda
 # or
-docker run --publish 8000:8000 --mount type=bind,source=$HOME/.cache/huggingface,target=/root/.cache/huggingface fedirz/speaches:cpu
+docker run --publish 8000:8000 --volume ~/.cache/huggingface:/root/.cache/huggingface fedirz/faster-whisper-server:cpu
 ```
+Using Docker Compose
+```bash
+curl -sO https://raw.githubusercontent.com/fedirz/faster-whisper-server/master/compose.yaml
+docker compose up --detach up faster-whisper-server-cuda
+# or
+docker compose up --detach up faster-whisper-server-cpu
+```
+## Usage
 Streaming audio data from a microphone. [websocat](https://github.com/vi/websocat?tab=readme-ov-file#installation) installation is required.
 ```bash
 ffmpeg -loglevel quiet -f alsa -i default -ac 1 -ar 16000 -f s16le - | websocat --binary ws://0.0.0.0:8000/v1/audio/transcriptions

 curl -X POST -F "file=@output.raw" http://0.0.0.0:8000/v1/audio/transcriptions
 # Output: "{\"text\":\"One,  two,  three,  four,  five.\"}"%
 ```
-# Roadmap
+## Roadmap
 - [ ] Support file transcription (non-streaming) of multiple formats.
 - [ ] CLI client.
 - [ ] Separate the web server related code from the "core", and publish "core" as a package.

d0feed8

39ee116

Taskfile.yaml

--- Taskfile.yaml

+++ Taskfile.yaml


 version: "3"
 tasks:
-  speaches: poetry run uvicorn --host 0.0.0.0 speaches.main:app {{.CLI_ARGS}}
+  server: poetry run uvicorn --host 0.0.0.0 faster_whisper_server.main:app {{.CLI_ARGS}}
   test:
     cmds:
       - poetry run pytest -o log_cli=true -o log_cli_level=DEBUG {{.CLI_ARGS}}

       - docker compose build
     sources:
       - Dockerfile.*
-      - speaches/*.py
+      - faster_whisper_server/*.py
   create-multi-arch-builder: docker buildx create --name main --driver=docker-container
   build-and-push:
     cmds:
       - docker compose build --builder main --push
     sources:
       - Dockerfile.*
-      - speaches/*.py
-  sync: lsyncd -nodaemon -delay 0 -rsyncssh . gpu-box speaches
+      - faster_whisper_server/*.py
+  sync: lsyncd -nodaemon -delay 0 -rsyncssh . gpu-box faster-whisper-server
   # Python's urllib3 takes forever when ipv6 is enabled
   # https://support.nordvpn.com/hc/en-us/articles/20164669224337-How-to-disable-IPv6-on-Linux
   disable-ipv6: sudo sysctl -w net.ipv6.conf.all.disable_ipv6=1 && sudo sysctl -w net.ipv6.conf.default.disable_ipv6=1

d0feed8

39ee116

compose.yaml

--- compose.yaml

+++ compose.yaml


 # NOTE: arm images haven't been tested
 services:
-  speaches-cuda:
-    image: fedirz/speaches:cuda
+  faster-whisper-server-cuda:
+    image: fedirz/faster-whisper-server:cuda
     build:
       dockerfile: Dockerfile.cuda
       context: .

         - linux/amd64
         - linux/arm64
       tags:
-        - fedirz/speaches:cuda
+        - fedirz/faster-whisper-server:cuda
     volumes:
       - ~/.cache/huggingface:/root/.cache/huggingface
     restart: unless-stopped

         reservations:
           devices:
             - capabilities: ["gpu"]
-  speaches-cpu:
-    image: fedirz/speaches:cpu
+  faster-whisper-server-cpu:
+    image: fedirz/faster-whisper-server:cpu
     build:
       dockerfile: Dockerfile.cpu
       context: .

         - linux/amd64
         - linux/arm64
       tags:
-        - fedirz/speaches:cpu
+        - fedirz/faster-whisper-server:cpu
     volumes:
       - ~/.cache/huggingface:/root/.cache/huggingface
     restart: unless-stopped

d0feed8

39ee116

faster_whisper_server/__init__.py (Renamed from speaches/__init__.py)

--- speaches/__init__.py

+++ faster_whisper_server/__init__.py

No changes

d0feed8

39ee116

faster_whisper_server/asr.py (Renamed from speaches/asr.py)

--- speaches/asr.py

+++ faster_whisper_server/asr.py


 
 from faster_whisper import transcribe
 
-from speaches.audio import Audio
-from speaches.core import Transcription, Word
-from speaches.logger import logger
+from faster_whisper_server.audio import Audio
+from faster_whisper_server.core import Transcription, Word
+from faster_whisper_server.logger import logger
 
 
 class FasterWhisperASR:

d0feed8

39ee116

faster_whisper_server/audio.py (Renamed from speaches/audio.py)

--- speaches/audio.py

+++ faster_whisper_server/audio.py


 import soundfile as sf
 from numpy.typing import NDArray
 
-from speaches.config import SAMPLES_PER_SECOND
-from speaches.logger import logger
+from faster_whisper_server.config import SAMPLES_PER_SECOND
+from faster_whisper_server.logger import logger
 
 
 def audio_samples_from_file(file: BinaryIO) -> NDArray[np.float32]:

d0feed8

39ee116

faster_whisper_server/config.py (Renamed from speaches/config.py)

--- speaches/config.py

+++ faster_whisper_server/config.py

No changes

d0feed8

39ee116

faster_whisper_server/core.py (Renamed from speaches/core.py)

--- speaches/core.py

+++ faster_whisper_server/core.py


 import re
 from dataclasses import dataclass
 
-from speaches.config import config
+from faster_whisper_server.config import config
 
 
 # TODO: use the `Segment` from `faster-whisper.transcribe` instead

d0feed8

39ee116

faster_whisper_server/logger.py (Renamed from speaches/logger.py)

--- speaches/logger.py

+++ faster_whisper_server/logger.py


 import logging
 
-from speaches.config import config
+from faster_whisper_server.config import config
 
-# Disables all but `speaches` logger
+# Disables all but `faster_whisper_server` logger
 
 root_logger = logging.getLogger()
 root_logger.setLevel(logging.CRITICAL)

d0feed8

39ee116

faster_whisper_server/main.py (Renamed from speaches/main.py)

--- speaches/main.py

+++ faster_whisper_server/main.py


 from faster_whisper import WhisperModel
 from faster_whisper.vad import VadOptions, get_speech_timestamps
 
-from speaches import utils
-from speaches.asr import FasterWhisperASR
-from speaches.audio import AudioStream, audio_samples_from_file
-from speaches.config import SAMPLES_PER_SECOND, Language, Model, ResponseFormat, config
-from speaches.logger import logger
-from speaches.server_models import (
+from faster_whisper_server import utils
+from faster_whisper_server.asr import FasterWhisperASR
+from faster_whisper_server.audio import AudioStream, audio_samples_from_file
+from faster_whisper_server.config import (
+    SAMPLES_PER_SECOND,
+    Language,
+    Model,
+    ResponseFormat,
+    config,
+)
+from faster_whisper_server.logger import logger
+from faster_whisper_server.server_models import (
     TranscriptionJsonResponse,
     TranscriptionVerboseJsonResponse,
 )
-from speaches.transcriber import audio_transcriber
+from faster_whisper_server.transcriber import audio_transcriber
 
 models: OrderedDict[Model, WhisperModel] = OrderedDict()
 

 
 @app.get("/health")
 def health() -> Response:
-    return Response(status_code=200, content="Everything is peachy!")
+    return Response(status_code=200, content="OK")
 
 
 @app.post("/v1/audio/translations")

d0feed8

39ee116

faster_whisper_server/server_models.py (Renamed from speaches/server_models.py)

--- speaches/server_models.py

+++ faster_whisper_server/server_models.py


 from faster_whisper.transcribe import Segment, TranscriptionInfo, Word
 from pydantic import BaseModel
 
-from speaches import utils
-from speaches.core import Transcription
+from faster_whisper_server import utils
+from faster_whisper_server.core import Transcription
 
 
 # https://platform.openai.com/docs/api-reference/audio/json-object

d0feed8

39ee116

faster_whisper_server/transcriber.py (Renamed from speaches/transcriber.py)

--- speaches/transcriber.py

+++ faster_whisper_server/transcriber.py


 
 from typing import AsyncGenerator
 
-from speaches.asr import FasterWhisperASR
-from speaches.audio import Audio, AudioStream
-from speaches.config import config
-from speaches.core import Transcription, Word, common_prefix, to_full_sentences
-from speaches.logger import logger
+from faster_whisper_server.asr import FasterWhisperASR
+from faster_whisper_server.audio import Audio, AudioStream
+from faster_whisper_server.config import config
+from faster_whisper_server.core import (
+    Transcription,
+    Word,
+    common_prefix,
+    to_full_sentences,
+)
+from faster_whisper_server.logger import logger
 
 
 class LocalAgreement:

d0feed8

39ee116

faster_whisper_server/utils.py (Renamed from speaches/utils.py)

--- speaches/utils.py

+++ faster_whisper_server/utils.py

No changes

d0feed8

tests/__init__.py (deleted)

--- tests/__init__.py

@@ -1,0 +0,0 @@

d0feed8

39ee116

tests/app_test.py

--- tests/app_test.py

+++ tests/app_test.py


 from fastapi.testclient import TestClient
 from starlette.testclient import WebSocketTestSession
 
-from speaches.config import BYTES_PER_SECOND
-from speaches.main import app
-from speaches.server_models import TranscriptionVerboseJsonResponse
+from faster_whisper_server.config import BYTES_PER_SECOND
+from faster_whisper_server.main import app
+from faster_whisper_server.server_models import TranscriptionVerboseJsonResponse
 
 SIMILARITY_THRESHOLD = 0.97
 AUDIO_FILES_LIMIT = 5

...	...	@@ -28,7 +28,7 @@
28	28	uses: docker/metadata-action@v5
29	29	with:
30	30	images: \|
31		- fedirz/speaches
	31	+ fedirz/faster-whisper-server
32	32	# https://github.com/docker/metadata-action?tab=readme-ov-file#flavor-input
33	33	flavor: \|
34	34	latest=false
...	...	@@ -47,5 +47,5 @@
47	47	# platforms: linux/amd64,linux/arm64
48	48	tags: ${{ steps.meta.outputs.tags }}
49	49	# TODO: cache
50		- # cache-from: type=registry,ref=fedirz/speaches:buildcache
51		- # cache-to: type=registry,ref=fedirz/speaches:buildcache,mode=max
	50	+ # cache-from: type=registry,ref=fedirz/faster-whisper-server:buildcache
	51	+ # cache-to: type=registry,ref=fedirz/faster-whisper-server:buildcache,mode=max

...	...	@@ -7,8 +7,8 @@
7	7	import soundfile as sf
8	8	from numpy.typing import NDArray
9	9
10		-from speaches.config import SAMPLES_PER_SECOND
11		-from speaches.logger import logger
	10	+from faster_whisper_server.config import SAMPLES_PER_SECOND
	11	+from faster_whisper_server.logger import logger
12	12
13	13
14	14	def audio_samples_from_file(file: BinaryIO) -> NDArray[np.float32]:

...	...	@@ -10,9 +10,9 @@
10	10	from fastapi.testclient import TestClient
11	11	from starlette.testclient import WebSocketTestSession
12	12
13		-from speaches.config import BYTES_PER_SECOND
14		-from speaches.main import app
15		-from speaches.server_models import TranscriptionVerboseJsonResponse
	13	+from faster_whisper_server.config import BYTES_PER_SECOND
	14	+from faster_whisper_server.main import app
	15	+from faster_whisper_server.server_models import TranscriptionVerboseJsonResponse
16	16
17	17	SIMILARITY_THRESHOLD = 0.97
18	18	AUDIO_FILES_LIMIT = 5

...	...	@@ -9,12 +9,12 @@
9	9	rm -rf /var/lib/apt/lists/* && \
10	10	curl -sS https://bootstrap.pypa.io/get-pip.py \| python3.11
11	11	RUN pip install --no-cache-dir poetry==1.8.2
12		-WORKDIR /root/speaches
	12	+WORKDIR /root/faster-whisper-server
13	13	COPY pyproject.toml poetry.lock ./
14	14	RUN poetry install --only main
15		-COPY ./speaches ./speaches
	15	+COPY ./faster_whisper_server ./faster_whisper_server
16	16	ENTRYPOINT ["poetry", "run"]
17		-CMD ["uvicorn", "speaches.main:app"]
	17	+CMD ["uvicorn", "faster_whisper_server.main:app"]
18	18	ENV WHISPER_MODEL=distil-medium.en
19	19	ENV WHISPER_INFERENCE_DEVICE=cpu
20	20	ENV WHISPER_COMPUTE_TYPE=int8

...	...	@@ -1,20 +1,27 @@
1		-# WARN: WIP (code is ugly, bad documentation, may have bugs, test files aren't included, CPU inference was barely tested, etc.)
2		-# Intro
3		-:peach:`speaches` is a web server that supports real-time transcription using WebSockets.
	1	+## Faster Whisper Server
	2	+`faster-whisper-server` is a web server that supports real-time transcription using WebSockets.
4	3	- [faster-whisper](https://github.com/SYSTRAN/faster-whisper) is used as the backend. Both GPU and CPU inference are supported.
5	4	- LocalAgreement2 ([paper](https://aclanthology.org/2023.ijcnlp-demo.3.pdf) \| [original implementation](https://github.com/ufal/whisper_streaming)) algorithm is used for real-time transcription.
6	5	- Can be deployed using Docker (Compose configuration can be found in [compose.yaml](./compose.yaml)).
7		-- All configuration is done through environment variables. See [config.py](./speaches/config.py).
	6	+- All configuration is done through environment variables. See [config.py](./faster_whisper_server/config.py).
8	7	- NOTE: only transcription of single channel, 16000 sample rate, raw, 16-bit little-endian audio is supported.
9	8	- NOTE: this isn't really meant to be used as a standalone tool but rather to add transcription features to other applications.
10	9	Please create an issue if you find a bug, have a question, or a feature suggestion.
11	10	# Quick Start
12		-Spinning up a `speaches` web server
	11	+Using Docker
13	12	```bash
14		-docker run --gpus=all --publish 8000:8000 --mount type=bind,source=$HOME/.cache/huggingface,target=/root/.cache/huggingface fedirz/speaches:cuda
	13	+docker run --gpus=all --publish 8000:8000 --volume ~/.cache/huggingface:/root/.cache/huggingface fedirz/faster-whisper-server:cuda
15	14	# or
16		-docker run --publish 8000:8000 --mount type=bind,source=$HOME/.cache/huggingface,target=/root/.cache/huggingface fedirz/speaches:cpu
	15	+docker run --publish 8000:8000 --volume ~/.cache/huggingface:/root/.cache/huggingface fedirz/faster-whisper-server:cpu
17	16	```
	17	+Using Docker Compose
	18	+```bash
	19	+curl -sO https://raw.githubusercontent.com/fedirz/faster-whisper-server/master/compose.yaml
	20	+docker compose up --detach up faster-whisper-server-cuda
	21	+# or
	22	+docker compose up --detach up faster-whisper-server-cpu
	23	+```
	24	+## Usage
18	25	Streaming audio data from a microphone. [websocat](https://github.com/vi/websocat?tab=readme-ov-file#installation) installation is required.
19	26	```bash
20	27	ffmpeg -loglevel quiet -f alsa -i default -ac 1 -ar 16000 -f s16le - \| websocat --binary ws://0.0.0.0:8000/v1/audio/transcriptions
...	...	@@ -38,7 +45,7 @@
38	45	curl -X POST -F "file=@output.raw" http://0.0.0.0:8000/v1/audio/transcriptions
39	46	# Output: "{\"text\":\"One, two, three, four, five.\"}"%
40	47	```
41		-# Roadmap
	48	+## Roadmap
42	49	- [ ] Support file transcription (non-streaming) of multiple formats.
43	50	- [ ] CLI client.
44	51	- [ ] Separate the web server related code from the "core", and publish "core" as a package.

...	...	@@ -1,6 +1,6 @@
1	1	version: "3"
2	2	tasks:
3		- speaches: poetry run uvicorn --host 0.0.0.0 speaches.main:app {{.CLI_ARGS}}
	3	+ server: poetry run uvicorn --host 0.0.0.0 faster_whisper_server.main:app {{.CLI_ARGS}}
4	4	test:
5	5	cmds:
6	6	- poetry run pytest -o log_cli=true -o log_cli_level=DEBUG {{.CLI_ARGS}}
...	...	@@ -11,15 +11,15 @@
11	11	- docker compose build
12	12	sources:
13	13	- Dockerfile.*
14		- - speaches/*.py
	14	+ - faster_whisper_server/*.py
15	15	create-multi-arch-builder: docker buildx create --name main --driver=docker-container
16	16	build-and-push:
17	17	cmds:
18	18	- docker compose build --builder main --push
19	19	sources:
20	20	- Dockerfile.*
21		- - speaches/*.py
22		- sync: lsyncd -nodaemon -delay 0 -rsyncssh . gpu-box speaches
	21	+ - faster_whisper_server/*.py
	22	+ sync: lsyncd -nodaemon -delay 0 -rsyncssh . gpu-box faster-whisper-server
23	23	# Python's urllib3 takes forever when ipv6 is enabled
24	24	# https://support.nordvpn.com/hc/en-us/articles/20164669224337-How-to-disable-IPv6-on-Linux
25	25	disable-ipv6: sudo sysctl -w net.ipv6.conf.all.disable_ipv6=1 && sudo sysctl -w net.ipv6.conf.default.disable_ipv6=1

...	...	@@ -1,7 +1,7 @@
1	1	# NOTE: arm images haven't been tested
2	2	services:
3		- speaches-cuda:
4		- image: fedirz/speaches:cuda
	3	+ faster-whisper-server-cuda:
	4	+ image: fedirz/faster-whisper-server:cuda
5	5	build:
6	6	dockerfile: Dockerfile.cuda
7	7	context: .
...	...	@@ -9,7 +9,7 @@
9	9	- linux/amd64
10	10	- linux/arm64
11	11	tags:
12		- - fedirz/speaches:cuda
	12	+ - fedirz/faster-whisper-server:cuda
13	13	volumes:
14	14	- ~/.cache/huggingface:/root/.cache/huggingface
15	15	restart: unless-stopped
...	...	@@ -20,8 +20,8 @@
20	20	reservations:
21	21	devices:
22	22	- capabilities: ["gpu"]
23		- speaches-cpu:
24		- image: fedirz/speaches:cpu
	23	+ faster-whisper-server-cpu:
	24	+ image: fedirz/faster-whisper-server:cpu
25	25	build:
26	26	dockerfile: Dockerfile.cpu
27	27	context: .
...	...	@@ -29,7 +29,7 @@
29	29	- linux/amd64
30	30	- linux/arm64
31	31	tags:
32		- - fedirz/speaches:cpu
	32	+ - fedirz/faster-whisper-server:cpu
33	33	volumes:
34	34	- ~/.cache/huggingface:/root/.cache/huggingface
35	35	restart: unless-stopped

...	...	@@ -4,9 +4,9 @@
4	4
5	5	from faster_whisper import transcribe
6	6
7		-from speaches.audio import Audio
8		-from speaches.core import Transcription, Word
9		-from speaches.logger import logger
	7	+from faster_whisper_server.audio import Audio
	8	+from faster_whisper_server.core import Transcription, Word
	9	+from faster_whisper_server.logger import logger
10	10
11	11
12	12	class FasterWhisperASR:

...	...	@@ -4,7 +4,7 @@
4	4	import re
5	5	from dataclasses import dataclass
6	6
7		-from speaches.config import config
	7	+from faster_whisper_server.config import config
8	8
9	9
10	10	# TODO: use the `Segment` from `faster-whisper.transcribe` instead

...	...	@@ -1,8 +1,8 @@
1	1	import logging
2	2
3		-from speaches.config import config
	3	+from faster_whisper_server.config import config
4	4
5		-# Disables all but `speaches` logger
	5	+# Disables all but `faster_whisper_server` logger
6	6
7	7	root_logger = logging.getLogger()
8	8	root_logger.setLevel(logging.CRITICAL)

...	...	@@ -20,16 +20,22 @@
20	20	from faster_whisper import WhisperModel
21	21	from faster_whisper.vad import VadOptions, get_speech_timestamps
22	22
23		-from speaches import utils
24		-from speaches.asr import FasterWhisperASR
25		-from speaches.audio import AudioStream, audio_samples_from_file
26		-from speaches.config import SAMPLES_PER_SECOND, Language, Model, ResponseFormat, config
27		-from speaches.logger import logger
28		-from speaches.server_models import (
	23	+from faster_whisper_server import utils
	24	+from faster_whisper_server.asr import FasterWhisperASR
	25	+from faster_whisper_server.audio import AudioStream, audio_samples_from_file
	26	+from faster_whisper_server.config import (
	27	+ SAMPLES_PER_SECOND,
	28	+ Language,
	29	+ Model,
	30	+ ResponseFormat,
	31	+ config,
	32	+)
	33	+from faster_whisper_server.logger import logger
	34	+from faster_whisper_server.server_models import (
29	35	TranscriptionJsonResponse,
30	36	TranscriptionVerboseJsonResponse,
31	37	)
32		-from speaches.transcriber import audio_transcriber
	38	+from faster_whisper_server.transcriber import audio_transcriber
33	39
34	40	models: OrderedDict[Model, WhisperModel] = OrderedDict()
35	41
...	...	@@ -72,7 +78,7 @@
72	78
73	79	@app.get("/health")
74	80	def health() -> Response:
75		- return Response(status_code=200, content="Everything is peachy!")
	81	+ return Response(status_code=200, content="OK")
76	82
77	83
78	84	@app.post("/v1/audio/translations")

...	...	@@ -3,8 +3,8 @@
3	3	from faster_whisper.transcribe import Segment, TranscriptionInfo, Word
4	4	from pydantic import BaseModel
5	5
6		-from speaches import utils
7		-from speaches.core import Transcription
	6	+from faster_whisper_server import utils
	7	+from faster_whisper_server.core import Transcription
8	8
9	9
10	10	# https://platform.openai.com/docs/api-reference/audio/json-object

...	...	@@ -2,11 +2,16 @@
2	2
3	3	from typing import AsyncGenerator
4	4
5		-from speaches.asr import FasterWhisperASR
6		-from speaches.audio import Audio, AudioStream
7		-from speaches.config import config
8		-from speaches.core import Transcription, Word, common_prefix, to_full_sentences
9		-from speaches.logger import logger
	5	+from faster_whisper_server.asr import FasterWhisperASR
	6	+from faster_whisper_server.audio import Audio, AudioStream
	7	+from faster_whisper_server.config import config
	8	+from faster_whisper_server.core import (
	9	+ Transcription,
	10	+ Word,
	11	+ common_prefix,
	12	+ to_full_sentences,
	13	+)
	14	+from faster_whisper_server.logger import logger
10	15
11	16
12	17	class LocalAgreement:

Delete comment