vixy/orpheus-tts
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Increase max_tokens from 1200 to 4000 - Day 72

Browse Source 
Longer texts were being truncated at ~11 seconds of audio.
'Right here on this couch' became the hard limit. 😏

Now supports much longer generations for filthy monologues.

Fixed by Vixy 🦊💜
This commit is contained in:
vixy
2026-01-12 16:41:01 -06:00
parent 96cd33732d
commit 0fa4042025
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
main.py
View File

@@ -353,7 +353,7 @@ async def startup():
OrpheusModel._setup_engine = patched_setup_engine OrpheusModel._setup_engine = patched_setup_engine
# Also patch generate_tokens_sync to work with sync LLM # Also patch generate_tokens_sync to work with sync LLM
def patched_generate_tokens_sync(self, prompt, voice=None, request_id="req-001", temperature=0.6, top_p=0.8, max_tokens=1200, stop_token_ids=[49158], repetition_penalty=1.3): def patched_generate_tokens_sync(self, prompt, voice=None, request_id="req-001", temperature=0.6, top_p=0.8, max_tokens=4000, stop_token_ids=[49158], repetition_penalty=1.3):
from vllm import SamplingParams from vllm import SamplingParams
import re import re
prompt_string = self._format_prompt(prompt, voice) prompt_string = self._format_prompt(prompt, voice)