chore(format): run black on main

ChatTTS/core.py

@@ -70,10 +70,15 @@ def download_models(
     ) -> Optional[str]:
         if source == "local":
             download_path = os.getcwd()
-            if not check_all_assets(Path(download_path), self.sha256_map, update=True) or force_redownload:
+            if (
+                not check_all_assets(Path(download_path), self.sha256_map, update=True)
+                or force_redownload
+            ):
                 with tempfile.TemporaryDirectory() as tmp:
                     download_all_assets(tmpdir=tmp)
-                if not check_all_assets(Path(download_path), self.sha256_map, update=False):
+                if not check_all_assets(
+                    Path(download_path), self.sha256_map, update=False
+                ):
                     self.logger.error(
                         "download to local path %s failed.", download_path
                     )
@@ -107,9 +112,7 @@ def download_models(
         elif source == "custom":
             self.logger.log(logging.INFO, f"try to load from local: {custom_path}")
             if not check_all_assets(Path(custom_path), self.sha256_map, update=False):
-                self.logger.error(
-                    "check models in custom path %s failed.", custom_path
-                )
+                self.logger.error("check models in custom path %s failed.", custom_path)
                 return None
             download_path = custom_path
 

ChatTTS/model/gpt.py

@@ -226,7 +226,9 @@ def _prepare_generation_inputs(
             # 2 - If the past_length is smaller than input_ids', then input_ids holds all input tokens. We can discard
             # input_ids based on the past_length.
             elif past_length < input_ids.shape[1]:
-                input_ids = input_ids.narrow(1, past_length, input_ids.size(1)-past_length)
+                input_ids = input_ids.narrow(
+                    1, past_length, input_ids.size(1) - past_length
+                )
             # 3 - Otherwise (past_length >= input_ids.shape[1]), let's assume input_ids only has unprocessed tokens.
 
             # If we are about to go beyond the maximum cache length, we need to crop the input attention mask.
@@ -235,14 +237,18 @@ def _prepare_generation_inputs(
                 and attention_mask is not None
                 and cache_length + input_ids.shape[1] > max_cache_length
             ):
-                attention_mask = attention_mask.narrow(1, -max_cache_length, max_cache_length)
+                attention_mask = attention_mask.narrow(
+                    1, -max_cache_length, max_cache_length
+                )
 
         if attention_mask is not None and position_ids is None:
             # create position_ids on the fly for batch generation
             position_ids = attention_mask.long().cumsum(-1) - 1
             position_ids.masked_fill_(attention_mask == 0, 1)
             if past_key_values:
-                position_ids = position_ids.narrow(1, -input_ids.shape[1], input_ids.shape[1])
+                position_ids = position_ids.narrow(
+                    1, -input_ids.shape[1], input_ids.shape[1]
+                )
 
         input_length = (
             position_ids.shape[-1] if position_ids is not None else input_ids.shape[-1]
@@ -360,7 +366,9 @@ def generate(
                 device=inputs_ids.device,
             )
             if attention_mask is not None:
-                attention_mask_cache.narrow(1, 0, attention_mask.shape[1]).copy_(attention_mask)
+                attention_mask_cache.narrow(1, 0, attention_mask.shape[1]).copy_(
+                    attention_mask
+                )
 
             with tqdm(
                 total=max_new_token,

ChatTTS/model/processors.py

@@ -22,7 +22,9 @@ def __call__(
             input_ids = input_ids.narrow(1, -self.past_window, self.past_window)
         freq = F.one_hot(input_ids, scores.size(1)).sum(1)
         if freq.size(0) > self.max_input_ids:
-            freq.narrow(0, self.max_input_ids, freq.size(0)-self.max_input_ids).zero_()
+            freq.narrow(
+                0, self.max_input_ids, freq.size(0) - self.max_input_ids
+            ).zero_()
         alpha = torch.pow(self.penalty, freq)
         scores = scores.contiguous()
         inp = scores.multiply(alpha)
@@ -32,6 +34,7 @@ def __call__(
         del inp, oth, scores, con, alpha
         return out
 
+
 def gen_logits(
     num_code: int,
     top_P=0.7,

examples/web/ex.py

@@ -1,4 +1,4 @@
-ex=[
+ex = [
     [
         "四川美食确实以辣闻名，但也有不辣的选择。比如甜水面、赖汤圆、蛋烘糕、叶儿粑等，这些小吃口味温和，甜而不腻，也很受欢迎。",
         0.3,

examples/web/funcs.py

@@ -108,7 +108,7 @@ def refine_text(
     has_interrupted = False
 
     if not refine_text_flag:
-        sleep(1) # to skip fast answer of loading mark
+        sleep(1)  # to skip fast answer of loading mark
         return text, *set_generate_buttons(
             generate_button, interrupt_button, is_reset=True
         )