Install from Hugging Face; auto-download weights

README.md

@@ -30,8 +30,9 @@ Hugin is a small, fast Swedish text-to-speech model that also reads English, bot
 ## Install and use
 
 ```bash
-git clone https://github.com/Shimmy/futhark-hugin.git && cd futhark-hugin
-pip install -e ".[prominence]" && python -c "import stanza; stanza.download('sv', processors='tokenize,pos')"
+pip install -U huggingface_hub
+hf download codah80/futhark-hugin --local-dir futhark-hugin
+pip install -e "./futhark-hugin[prominence]" && python -c "import stanza; stanza.download('sv', processors='tokenize,pos')"
 python -m hugin "Hej och välkomna! Klockan är 18:30." -o hej.wav
 ```
 

docs/DETAILS.md

@@ -47,9 +47,10 @@ Regenerate them with `python tools/make_samples.py` (writes `samples/*.mp3`, nee
 ## Install
 
 ```bash
-git clone git@github.com:Shimmy/futhark-hugin.git
+pip install -U huggingface_hub
+hf download codah80/futhark-hugin --local-dir futhark-hugin   # or: git lfs install && git clone https://huggingface.co/codah80/futhark-hugin
 cd futhark-hugin
-pip install -e .            # torch, numpy, scipy, soundfile, num2words
+pip install -e .            # torch, numpy, scipy, soundfile, num2words, huggingface_hub
 pip install -e ".[server]"  # optional: the small HTTP server (fastapi, uvicorn)
 pip install -e ".[finetune]" # optional: hugin-finetune, to make your own voice (torchaudio, transformers)
 pip install -e ".[prominence]" && python -c "import stanza; stanza.download('sv', processors='tokenize,pos')"
@@ -58,7 +59,7 @@ pip install -e ".[prominence]" && python -c "import stanza; stanza.download('sv'
 
 **Prominence is optional.** With Stanza installed Hugin de-accents function words and keeps the stress on verb particles ("hon **tog** inte **upp** den"), which fixes most phrasing errors. Without Stanza it runs as before (the particle-verb list and the other rules still apply, but not the part-of-speech step). `prominence="auto"` (default) uses Stanza if it is installed, `prominence=False` / `--no-prominence` turns it off. Stanza looks for its model in `~/stanza_resources` or `$STANZA_RESOURCES_DIR`; nothing is fetched at runtime.
 
-The weights live in `weights/`, about 150 MB stored as plain git files. An editable install finds them on its own. If you install the package somewhere else, set `HUGIN_WEIGHTS=/path/to/weights` or pass `path=` to `from_pretrained`.
+The weights live in `weights/` (about 180 MB). An editable install of the downloaded repository finds them on its own. If the package is installed without them, `from_pretrained()` downloads `weights/` from the Hugging Face repository once (cached by huggingface_hub; set `HUGIN_REPO` to use another copy). To use a local folder instead, set `HUGIN_WEIGHTS=/path/to/weights` or pass `path=` to `from_pretrained`.
 
 ## Quick start
 

hugin/synth.py

@@ -35,10 +35,18 @@ PUNCT = set(",.!?;:…\"'()–-")
 HPF_HZ = 80.0
 
 
+HF_REPO = os.environ.get("HUGIN_REPO", "codah80/futhark-hugin")
+
+
 def default_weights_dir():
-    """$HUGIN_WEIGHTS, else the weights/ folder next to the package (repo checkout / editable install)."""
+    """$HUGIN_WEIGHTS, else the weights/ folder next to the package (a downloaded copy of the Hugging Face repo, editable install),
+    else download weights/ from the Hugging Face repo once (cached by huggingface_hub, about 180 MB)."""
     env = os.environ.get("HUGIN_WEIGHTS")
-    return Path(env) if env else Path(__file__).resolve().parent.parent / "weights"
+    if env: return Path(env)
+    local = Path(__file__).resolve().parent.parent / "weights"
+    if (local / "hugin-v0.8-b1.fp16.pt").exists(): return local
+    from huggingface_hub import snapshot_download
+    return Path(snapshot_download(HF_REPO, allow_patterns=["weights/*", "weights/**"])) / "weights"
 
 
 def _fp32(sd):

pyproject.toml

@@ -12,6 +12,7 @@ requires-python = ">=3.9"
 authors = [{ name = "Conny Dahlgren", email = "conny.dahlgren@devsec.se" }]
 dependencies = [
     "torch>=2.1",
+    "huggingface_hub>=0.25",
     "numpy",
     "scipy",
     "soundfile",