Mirroristas/whisper
2
0
Fork 0
mirror of https://github.com/openai/whisper.git synced 2025-11-28 08:11:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

Omit space prefix in initial_prompt for spaceless languages.

Browse Source
This commit is contained in:
Ryan Heise 2024-03-31 13:11:45 +11:00
parent ba3f3cd54b
commit 21999e1702
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
whisper/transcribe.py
View File

@ -228,7 +228,8 @@ def transcribe(
prompt_reset_since = 0
if initial_prompt is not None:
initial_prompt_tokens = tokenizer.encode(" " + initial_prompt.strip())
space = "" if language in {"zh", "ja", "th", "lo", "my", "yue"} else " "
initial_prompt_tokens = tokenizer.encode(space + initial_prompt.strip())
all_tokens.extend(initial_prompt_tokens)
else:
initial_prompt_tokens = []