tipi/kyutai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Remove unused HF repo arguments and reset audio output data in TTS server script

Browse Source
This commit is contained in:
Václav Volhejn 2025-07-03 09:42:27 +02:00
parent df36dfd918
commit 849302fb3f
1 changed files with 1 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
scripts/tts_rust_server.py
View File

@ -24,7 +24,6 @@ import websockets
SAMPLE_RATE = 24000 SAMPLE_RATE = 24000
TTS_TEXT = "Hello, this is a test of the moshi text to speech system, this should result in some nicely sounding generated voice." TTS_TEXT = "Hello, this is a test of the moshi text to speech system, this should result in some nicely sounding generated voice."
DEFAULT_DSM_TTS_REPO = "kyutai/tts-1.6b-en_fr"
DEFAULT_DSM_TTS_VOICE_REPO = "kyutai/tts-voices" DEFAULT_DSM_TTS_VOICE_REPO = "kyutai/tts-voices"
AUTH_TOKEN = "public_token" AUTH_TOKEN = "public_token"
@ -64,6 +63,7 @@ async def output_audio(out: str, output_queue: asyncio.Queue[np.ndarray | None])
outdata[:, 0] = pcm_data outdata[:, 0] = pcm_data
else: else:
should_exit = True should_exit = True
outdata[:] = 0
except asyncio.QueueEmpty: except asyncio.QueueEmpty:
outdata[:] = 0 outdata[:] = 0
@ -95,17 +95,6 @@ async def websocket_client():
parser.add_argument( parser.add_argument(
"out", type=str, help="Output file to generate, use - for playing the audio" "out", type=str, help="Output file to generate, use - for playing the audio"
) )
parser.add_argument(
"--hf-repo",
type=str,
default=DEFAULT_DSM_TTS_REPO,
help="HF repo in which to look for the pretrained models.",
)
parser.add_argument(
"--voice-repo",
default=DEFAULT_DSM_TTS_VOICE_REPO,
help="HF repo in which to look for pre-computed voice embeddings.",
)
parser.add_argument( parser.add_argument(
"--voice", "--voice",
default="expresso/ex03-ex01_happy_001_channel1_334s.wav", default="expresso/ex03-ex01_happy_001_channel1_334s.wav",