Skip to content

SOTA Open Source TTS

Fix dockerfile for `pyaudio` (#623)

spicysama23fa4d7e3890
diff --git a/.github/workflows/build-docker-image.yml b/.github/workflows/build-docker-image.yml
index 6a2b7ef..28cb7e7 100644
--- a/.github/workflows/build-docker-image.yml
+++ b/.github/workflows/build-docker-image.yml
@@ -5,7 +5,7 @@ on:
branches:
- main
tags:
- - 'v*'
+ - "v*"
jobs:
build:
diff --git a/dockerfile b/dockerfile
index f22bea8..8cd7834 100644
--- a/dockerfile
+++ b/dockerfile
@@ -18,6 +18,10 @@ ARG DEPENDENCIES=" \
libsox-dev \
build-essential \
cmake \
+ libasound-dev \
+ portaudio19-dev \
+ libportaudio2 \
+ libportaudiocpp0 \
ffmpeg"
RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
diff --git a/dockerfile.dev b/dockerfile.dev
index 2d07296..ac5d18f 100644
--- a/dockerfile.dev
+++ b/dockerfile.dev
@@ -17,6 +17,10 @@ ARG TOOLS=" \
openssh-server \
sudo \
protobuf-compiler \
+ libasound-dev \
+ portaudio19-dev \
+ libportaudio2 \
+ libportaudiocpp0 \
cmake"
RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
diff --git a/tools/msgpack_api.py b/tools/msgpack_api.py
index 56fbec3..77e6861 100644
--- a/tools/msgpack_api.py
+++ b/tools/msgpack_api.py
@@ -1,8 +1,14 @@
+import os
+from argparse import ArgumentParser
+from pathlib import Path
+
import httpx
import ormsgpack
from tools.commons import ServeReferenceAudio, ServeTTSRequest
+api_key = os.environ.get("FISH_API_KEY", "YOUR_API_KEY")
+
def audio_request():
# priority: ref_id > references
@@ -18,6 +24,8 @@ def audio_request():
streaming=True,
)
+ api_key = os.environ.get("FISH_API_KEY", "YOUR_API_KEY")
+
with (
httpx.Client() as client,
open("hello.wav", "wb") as f,
@@ -27,7 +35,7 @@ def audio_request():
"http://127.0.0.1:8080/v1/tts",
content=ormsgpack.packb(request, option=ormsgpack.OPT_SERIALIZE_PYDANTIC),
headers={
- "authorization": "Bearer YOUR_API_KEY",
+ "authorization": f"Bearer {api_key}",
"content-type": "application/msgpack",
},
timeout=None,
@@ -36,11 +44,11 @@ def audio_request():
f.write(chunk)
-def asr_request():
+def asr_request(audio_path: Path):
# Read the audio file
with open(
- r"D:\PythonProject\fish-speech\.cache\test_audios\prompts\2648200402409733590.wav",
+ str(audio_path),
"rb",
) as audio_file:
audio_data = audio_file.read()
@@ -57,7 +65,7 @@ def asr_request():
response = client.post(
"https://api.fish.audio/v1/asr",
headers={
- "Authorization": "Bearer 8eda4aeed2bc4aec9489b3efad003799",
+ "Authorization": f"Bearer {api_key}",
"Content-Type": "application/msgpack",
},
content=ormsgpack.packb(request_data),
@@ -74,5 +82,14 @@ def asr_request():
print(f"Start time: {segment['start']}, End time: {segment['end']}")
+def parse_args():
+ parser = ArgumentParser()
+ parser.add_argument("--audio_path", type=Path, default="audio/ref/trump.mp3")
+
+ return parser.parse_args()
+
+
if __name__ == "__main__":
- asr_request()
+ args = parse_args()
+
+ asr_request(args.audio_path)