Add speed control for inference (#3214)

* Add speed control for inference

* Fix XTTS tests

* Add speed control tests

TTS/tts/models/xtts.py

@@ -530,8 +530,10 @@ def inference(
         top_p=0.85,
         do_sample=True,
         num_beams=1,
+        speed=1.0,
         **hf_generate_kwargs,
     ):
+        length_scale = 1.0 / max(speed, 0.05)
         text = text.strip().lower()
         text_tokens = torch.IntTensor(self.tokenizer.encode(text, lang=language)).unsqueeze(0).to(self.device)
 
@@ -584,6 +586,13 @@ def inference(
                     gpt_latents = gpt_latents[:, :k]
                     break
 
+            if length_scale != 1.0:
+                gpt_latents = F.interpolate(
+                    gpt_latents.transpose(1, 2),
+                    scale_factor=length_scale,
+                    mode="linear"
+                ).transpose(1, 2)
+
             wav = self.hifigan_decoder(gpt_latents, g=speaker_embedding)
 
         return {
@@ -634,8 +643,10 @@ def inference_stream(
         top_k=50,
         top_p=0.85,
         do_sample=True,
+        speed=1.0,
         **hf_generate_kwargs,
     ):
+        length_scale = 1.0 / max(speed, 0.05)
         text = text.strip().lower()
         text_tokens = torch.IntTensor(self.tokenizer.encode(text, lang=language)).unsqueeze(0).to(self.device)
 
@@ -674,6 +685,12 @@ def inference_stream(
 
             if is_end or (stream_chunk_size > 0 and len(last_tokens) >= stream_chunk_size):
                 gpt_latents = torch.cat(all_latents, dim=0)[None, :]
+                if length_scale != 1.0:
+                    gpt_latents = F.interpolate(
+                        gpt_latents.transpose(1, 2),
+                        scale_factor=length_scale,
+                        mode="linear"
+                    ).transpose(1, 2)
                 wav_gen = self.hifigan_decoder(gpt_latents, g=speaker_embedding.to(self.device))
                 wav_chunk, wav_gen_prev, wav_overlap = self.handle_chunks(
                     wav_gen.squeeze(), wav_gen_prev, wav_overlap, overlap_wav_len

tests/zoo_tests/test_models.py

@@ -111,7 +111,7 @@ def test_xtts_streaming():
     model.to(torch.device("cuda" if torch.cuda.is_available() else "cpu"))
 
     print("Computing speaker latents...")
-    gpt_cond_latent, _, speaker_embedding = model.get_conditioning_latents(audio_path=speaker_wav)
+    gpt_cond_latent, speaker_embedding = model.get_conditioning_latents(audio_path=speaker_wav)
 
     print("Inference...")
     chunks = model.inference_stream(
@@ -139,7 +139,7 @@ def test_xtts_v2():
             "yes | "
             f"tts --model_name  tts_models/multilingual/multi-dataset/xtts_v2 "
             f'--text "This is an example." --out_path "{output_path}" --progress_bar False --use_cuda True '
-            f'--speaker_wav "{speaker_wav}" "{speaker_wav_2}"  "--language_idx "en"'
+            f'--speaker_wav "{speaker_wav}" "{speaker_wav_2}"  --language_idx "en"'
         )
     else:
         run_cli(
@@ -164,7 +164,7 @@ def test_xtts_v2_streaming():
     model.to(torch.device("cuda" if torch.cuda.is_available() else "cpu"))
 
     print("Computing speaker latents...")
-    gpt_cond_latent, _, speaker_embedding = model.get_conditioning_latents(audio_path=speaker_wav)
+    gpt_cond_latent, speaker_embedding = model.get_conditioning_latents(audio_path=speaker_wav)
 
     print("Inference...")
     chunks = model.inference_stream(
@@ -179,6 +179,34 @@ def test_xtts_v2_streaming():
             assert chunk.shape[-1] > 5000
         wav_chuncks.append(chunk)
     assert len(wav_chuncks) > 1
+    normal_len = sum([len(chunk) for chunk in wav_chuncks])
+
+    chunks = model.inference_stream(
+        "It took me quite a long time to develop a voice and now that I have it I am not going to be silent.",
+        "en",
+        gpt_cond_latent,
+        speaker_embedding,
+        speed=1.5
+    )
+    wav_chuncks = []
+    for i, chunk in enumerate(chunks):
+        wav_chuncks.append(chunk)
+    fast_len = sum([len(chunk) for chunk in wav_chuncks])
+
+    chunks = model.inference_stream(
+        "It took me quite a long time to develop a voice and now that I have it I am not going to be silent.",
+        "en",
+        gpt_cond_latent,
+        speaker_embedding,
+        speed=0.66
+    )
+    wav_chuncks = []
+    for i, chunk in enumerate(chunks):
+        wav_chuncks.append(chunk)
+    slow_len = sum([len(chunk) for chunk in wav_chuncks])
+
+    assert slow_len > normal_len
+    assert normal_len > fast_len
 
 
 def test_tortoise():