V1.1 package (#177)

* Fix button height

* Streaming support

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Convert to 1 channel

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Fix Conversion bug

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Fix target path

* Add checkpoint selection

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Fix gpup decorator

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Add link for labeler

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Localize labeler

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Add LoRA llama config

* Allow download stream audio

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* asr

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Add cache auto recycling

* 多打了一个字母

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Check 'compile' avaliable

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Add vits-decoder UI support & Fix bugs

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Configurable audio length, i18n

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Label Exception

* Add workflow

* Update build-windows-package.yml

* Update build-windows-package.yml

* Fix workflow: models remove

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Push to HF

* Modify Workflow

* Fix

* Free from Zipping

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

.github/workflows/build-windows-package.yml

@@ -36,9 +36,32 @@ jobs:
           mv fish-speech-zip/fish-speech/fishenv fish-speech-zip/fish-speech/ffmpeg.exe \
           fish-speech-zip/fish-speech/checkpoints fish-speech-zip/fish-speech/.cache ./fish-speech
           rm -rf fish-speech-zip
+          rm ./fish-speech/checkpoints/text2semantic-sft-large-v1-4k.pth
+          rm ./fish-speech/checkpoints/text2semantic-pretrain-medium-2k-v1.pth
           huggingface-cli download fishaudio/fish-speech-1 ffprobe.exe \
           --local-dir ./fish-speech --local-dir-use-symlinks False
+          huggingface-cli download fishaudio/fish-speech-1 text2semantic-sft-large-v1.1-4k.pth \
+          --local-dir ./fish-speech/checkpoints --local-dir-use-symlinks False
+          huggingface-cli download fishaudio/fish-speech-1 text2semantic-sft-medium-v1.1-4k.pth \
+          --local-dir ./fish-speech/checkpoints --local-dir-use-symlinks False
+          huggingface-cli download fishaudio/fish-speech-1 vits_decoder_v1.1.ckpt \
+          --local-dir ./fish-speech/checkpoints --local-dir-use-symlinks False
+
       - uses: actions/upload-artifact@v4
         with:
           name: fish-speech-main-${{ github.run_id }}
           path: ./fish-speech
+
+      - name: Check and Compress fish-speech
+        run: |
+          if [ ! -f fish-speech-main-${{ github.run_id }}.zip ]; then
+            zip -qr fish-speech-main-${{ github.run_id }}.zip ./fish-speech
+          else
+            echo "Zip file already exists, skipping compression."
+          fi
+
+      - name: Upload to Hugging Face
+        env:
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
+        run: |
+          huggingface-cli upload SpicyqSama007/fish-speech-packed ./fish-speech-main-${{ github.run_id }}.zip fish-speech-main-${{ github.run_id }}.zip

fish_speech/configs/text2semantic_finetune.yaml

@@ -5,14 +5,14 @@ defaults:
 
 project: text2semantic_finetune_dual_ar
 max_length: 2048
-ckpt_path: checkpoints/text2semantic-sft-medium-v1-4k.pth
+ckpt_path: checkpoints/text2semantic-sft-medium-v1.1-4k.pth
 resume_weights_only: true
 
 # Lightning Trainer
 trainer:
   accumulate_grad_batches: 1
   gradient_clip_val: 1.0
-  gradient_clip_algorithm: 'norm'
+  gradient_clip_algorithm: "norm"
   max_steps: 1000
   precision: bf16-true
   limit_val_batches: 10

tools/webui.py

@@ -148,8 +148,12 @@ def inference(
             yield (fake_audios * 32768).astype(np.int16).tobytes(), None, None
 
     if len(segments) == 0:
-        yield None, None, build_html_error_message(
-            i18n("No audio generated, please check the input text.")
+        return (
+            None,
+            None,
+            build_html_error_message(
+                i18n("No audio generated, please check the input text.")
+            ),
         )
 
     # No matter streaming or not, we need to return the final audio