Add ClearVoice MossFormer2 denoiser backend and static gap trimming

1 files changed, 5 insertions, 3 deletions

diff --git a/producer.sh b/producer.sh
index f2dc6be..d2a139f 100755
--- a/producer.sh
+++ b/producer.sh
@@ -7,9 +7,10 @@
 #
 # Trust boundary for downloads: official PyPI (uv, application wheels) and
 # Astral's managed CPython distribution. AI setup additionally trusts
-# download.pytorch.org and the pinned DeepFilterNet3 model host. Host
-# package-manager settings (extra indexes, targets, constraints, config
-# files) cannot add to or redirect these sources.
+# download.pytorch.org, the pinned DeepFilterNet3 model host, and Hugging Face
+# (ClearVoice MossFormer2_SE_48K weights). Host package-manager settings
+# (extra indexes, targets, constraints, config files) cannot add to or
+# redirect these sources.
 #
 # First bootstrap needs an existing `uv`, or a Python 3 with venv support.
 # Later runs, including moving or renaming this whole directory, are handled
@@ -44,6 +45,7 @@ Usage: ./producer.sh COMMAND [options], or simply:
   config                     Print resolved settings (TOML, or JSON with --json)
   doctor                     Check FFmpeg, data directory and CUDA driver
   setup --device cpu|cuda    Install and self-test the AI backend
+                             (--denoiser clearvoice for the MossFormer2 backend)
 
 Outputs are written beside each input by default. Run
 './producer.sh COMMAND --help' for the options of a command.