From 849302fb3fad6619803926e95255de3d24d1588a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?V=C3=A1clav=20Volhejn?= Date: Thu, 3 Jul 2025 09:42:27 +0200 Subject: [PATCH] Remove unused HF repo arguments and reset audio output data in TTS server script --- scripts/tts_rust_server.py | 13 +------------ 1 file changed, 1 insertion(+), 12 deletions(-) diff --git a/scripts/tts_rust_server.py b/scripts/tts_rust_server.py index 8a18439..9b67dc8 100644 --- a/scripts/tts_rust_server.py +++ b/scripts/tts_rust_server.py @@ -24,7 +24,6 @@ import websockets 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." -DEFAULT_DSM_TTS_REPO = "kyutai/tts-1.6b-en_fr" DEFAULT_DSM_TTS_VOICE_REPO = "kyutai/tts-voices" 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 else: should_exit = True + outdata[:] = 0 except asyncio.QueueEmpty: outdata[:] = 0 @@ -95,17 +95,6 @@ async def websocket_client(): parser.add_argument( "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( "--voice", default="expresso/ex03-ex01_happy_001_channel1_334s.wav",