Update app.py

app.py

@@ -56,36 +56,33 @@ from omegaconf import OmegaConf
 import torchaudio
 from torchaudio.transforms import Resample
 import soundfile as sf
-
 from tqdm import tqdm
 from einops import rearrange
 from codecmanipulator import CodecManipulator
 from mmtokenizer import _MMSentencePieceTokenizer
 from transformers import AutoTokenizer, AutoModelForCausalLM, LogitsProcessor, LogitsProcessorList
-import glob
-import time
-import copy
-from collections import Counter
 from models.soundstream_hubert_new import SoundStream
 from vocoder import build_codec_model, process_audio
 from post_process_audio import replace_low_freq_with_energy_matched
 
+# Install flash attention
+print("Installing flash-attn...")
+subprocess.run(
+    "pip install flash-attn --no-build-isolation",
+    env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
+    shell=True,
+)
+
+# Initialize device
 device = "cuda:0"
 
+# Load models once and reuse
+print("Loading models...")
 model = AutoModelForCausalLM.from_pretrained(
     "m-a-p/YuE-s1-7B-anneal-en-cot",
     torch_dtype=torch.float16,
-    attn_implementation="flash_attention_2",  # To enable flashattn, you have to install flash-attn
-).to(device)
-# assistant_model = AutoModelForCausalLM.from_pretrained(
-#     "m-a-p/YuE-s2-1B-general",
-#     torch_dtype=torch.float16,
-#     attn_implementation="flash_attention_2",  # To enable flashattn, you have to install flash-attn
-# ).to(device)
-# assistant_model = torch.compile(assistant_model)
-# model = torch.compile(model)
-# assistant_model.eval()
-model.eval()
+    attn_implementation="flash_attention_2",
+).to(device).eval()
 
 basic_model_config = './xcodec_mini_infer/final_ckpt/config.yaml'
 resume_path = './xcodec_mini_infer/final_ckpt/ckpt_00360000.pth'
@@ -93,308 +90,130 @@ config_path = './xcodec_mini_infer/decoders/config.yaml'
 vocal_decoder_path = './xcodec_mini_infer/decoders/decoder_131000.pth'
 inst_decoder_path = './xcodec_mini_infer/decoders/decoder_151000.pth'
 
-mmtokenizer = _MMSentencePieceTokenizer("./mm_tokenizer_v0.2_hf/tokenizer.model")
-
-codectool = CodecManipulator("xcodec", 0, 1)
-model_config = OmegaConf.load(basic_model_config)
 # Load codec model
+model_config = OmegaConf.load(basic_model_config)
 codec_model = eval(model_config.generator.name)(**model_config.generator.config).to(device)
-parameter_dict = torch.load(resume_path, map_location='cpu')
-codec_model.load_state_dict(parameter_dict['codec_model'])
-# codec_model = torch.compile(codec_model)
+codec_model.load_state_dict(torch.load(resume_path, map_location='cpu')['codec_model'])
 codec_model.eval()
 
 # Preload and compile vocoders
 vocal_decoder, inst_decoder = build_codec_model(config_path, vocal_decoder_path, inst_decoder_path)
-vocal_decoder.to(device)
-inst_decoder.to(device)
-# vocal_decoder = torch.compile(vocal_decoder)
-# inst_decoder = torch.compile(inst_decoder)
-vocal_decoder.eval()
-inst_decoder.eval()
-
-
-def generate_music(
-        max_new_tokens=5,
-        run_n_segments=2,
-        genre_txt=None,
-        lyrics_txt=None,
-        use_audio_prompt=False,
-        audio_prompt_path="",
-        prompt_start_time=0.0,
-        prompt_end_time=30.0,
-        cuda_idx=0,
-        rescale=False,
-):
-    if use_audio_prompt and not audio_prompt_path:
-        raise FileNotFoundError("Please offer audio prompt filepath using '--audio_prompt_path', when you enable 'use_audio_prompt'!")
-    cuda_idx = cuda_idx
-    max_new_tokens = max_new_tokens * 100
-    
-    with tempfile.TemporaryDirectory() as output_dir:
-        stage1_output_dir = os.path.join(output_dir, f"stage1")
-        os.makedirs(stage1_output_dir, exist_ok=True)
-
-        class BlockTokenRangeProcessor(LogitsProcessor):
-            def __init__(self, start_id, end_id):
-                self.blocked_token_ids = list(range(start_id, end_id))
-
-            def __call__(self, input_ids, scores):
-                scores[:, self.blocked_token_ids] = -float("inf")
-                return scores
-
-        def load_audio_mono(filepath, sampling_rate=16000):
-            audio, sr = torchaudio.load(filepath)
-            # Convert to mono
-            audio = torch.mean(audio, dim=0, keepdim=True)
-            # Resample if needed
-            if sr != sampling_rate:
-                resampler = Resample(orig_freq=sr, new_freq=sampling_rate)
-                audio = resampler(audio)
-            return audio
-
-        def split_lyrics(lyrics: str):
-            pattern = r"\[(\w+)\](.*?)\n(?=\[|\Z)"
-            segments = re.findall(pattern, lyrics, re.DOTALL)
-            structured_lyrics = [f"[{seg[0]}]\n{seg[1].strip()}\n\n" for seg in segments]
-            return structured_lyrics
-
-        # Call the function and print the result
-        stage1_output_set = []
-
-        genres = genre_txt.strip()
-        lyrics = split_lyrics(lyrics_txt + "\n")
-        # intruction
-        full_lyrics = "\n".join(lyrics)
-        prompt_texts = [f"Generate music from the given lyrics segment by segment.\n[Genre] {genres}\n{full_lyrics}"]
-        prompt_texts += lyrics
-
-        random_id = uuid.uuid4()
-        output_seq = None
-        # Here is suggested decoding config
-        top_p = 0.93
-        temperature = 1.0
-        repetition_penalty = 1.2
-        # special tokens
-        start_of_segment = mmtokenizer.tokenize('[start_of_segment]')
-        end_of_segment = mmtokenizer.tokenize('[end_of_segment]')
-
-        raw_output = None
-
-        # Format text prompt
-        run_n_segments = min(run_n_segments + 1, len(lyrics))
-
-        print(list(enumerate(tqdm(prompt_texts[:run_n_segments]))))
-
-        for i, p in enumerate(tqdm(prompt_texts[:run_n_segments])):
-            section_text = p.replace('[start_of_segment]', '').replace('[end_of_segment]', '')
-            guidance_scale = 1.5 if i <= 1 else 1.2
-            if i == 0:
-                continue
-            if i == 1:
-                if use_audio_prompt:
-                    audio_prompt = load_audio_mono(audio_prompt_path)
-                    audio_prompt.unsqueeze_(0)
-                    with torch.no_grad():
-                        raw_codes = codec_model.encode(audio_prompt.to(device), target_bw=0.5)
-                    raw_codes = raw_codes.transpose(0, 1)
-                    raw_codes = raw_codes.cpu().numpy().astype(np.int16)
-                    # Format audio prompt
-                    code_ids = codectool.npy2ids(raw_codes[0])
-                    audio_prompt_codec = code_ids[int(prompt_start_time * 50): int(prompt_end_time * 50)]  # 50 is tps of xcodec
-                    audio_prompt_codec_ids = [mmtokenizer.soa] + codectool.sep_ids + audio_prompt_codec + [
-                        mmtokenizer.eoa]
-                    sentence_ids = mmtokenizer.tokenize("[start_of_reference]") + audio_prompt_codec_ids + mmtokenizer.tokenize(
-                        "[end_of_reference]")
-                    head_id = mmtokenizer.tokenize(prompt_texts[0]) + sentence_ids
-                else:
-                    head_id = mmtokenizer.tokenize(prompt_texts[0])
-                prompt_ids = head_id + start_of_segment + mmtokenizer.tokenize(section_text) + [mmtokenizer.soa] + codectool.sep_ids
-            else:
-                prompt_ids = end_of_segment + start_of_segment + mmtokenizer.tokenize(section_text) + [mmtokenizer.soa] + codectool.sep_ids
-
-            prompt_ids = torch.as_tensor(prompt_ids).unsqueeze(0).to(device)
-            input_ids = torch.cat([raw_output, prompt_ids], dim=1) if i > 1 else prompt_ids
-            # Use window slicing in case output sequence exceeds the context of model
-            max_context = 16384 - max_new_tokens - 1
-            if input_ids.shape[-1] > max_context:
-                print(
-                    f'Section {i}: output length {input_ids.shape[-1]} exceeding context length {max_context}, now using the last {max_context} tokens.')
-                input_ids = input_ids[:, -(max_context):]
-            with torch.inference_mode(), torch.autocast(device_type='cuda', dtype=torch.float16):
-                output_seq = model.generate(
-                    input_ids=input_ids,
-                    max_new_tokens=max_new_tokens,
-                    min_new_tokens=100,
-                    do_sample=True,
-                    top_p=top_p,
-                    temperature=temperature,
-                    repetition_penalty=repetition_penalty,
-                    eos_token_id=mmtokenizer.eoa,
-                    pad_token_id=mmtokenizer.eoa,
-                    logits_processor=LogitsProcessorList([BlockTokenRangeProcessor(0, 32002), BlockTokenRangeProcessor(32016, 32016)]),
-                    guidance_scale=guidance_scale,
-                    use_cache=True,
-                    top_k=50,
-                    num_beams=1
-                )
-                if output_seq[0][-1].item() != mmtokenizer.eoa:
-                    tensor_eoa = torch.as_tensor([[mmtokenizer.eoa]]).to(model.device)
-                    output_seq = torch.cat((output_seq, tensor_eoa), dim=1)
-            if i > 1:
-                raw_output = torch.cat([raw_output, prompt_ids, output_seq[:, input_ids.shape[-1]:]], dim=1)
-            else:
-                raw_output = output_seq
-            print(len(raw_output))
-
-        # save raw output and check sanity
-        ids = raw_output[0].cpu().numpy()
-        soa_idx = np.where(ids == mmtokenizer.soa)[0].tolist()
-        eoa_idx = np.where(ids == mmtokenizer.eoa)[0].tolist()
-        if len(soa_idx) != len(eoa_idx):
-            raise ValueError(f'invalid pairs of soa and eoa, Num of soa: {len(soa_idx)}, Num of eoa: {len(eoa_idx)}')
-
-        vocals = []
-        instrumentals = []
-        range_begin = 1 if use_audio_prompt else 0
-        for i in range(range_begin, len(soa_idx)):
-            codec_ids = ids[soa_idx[i] + 1:eoa_idx[i]]
-            if codec_ids[0] == 32016:
-                codec_ids = codec_ids[1:]
-            codec_ids = codec_ids[:2 * (codec_ids.shape[0] // 2)]
-            vocals_ids = codectool.ids2npy(rearrange(codec_ids, "(n b) -> b n", b=2)[0])
-            vocals.append(vocals_ids)
-            instrumentals_ids = codectool.ids2npy(rearrange(codec_ids, "(n b) -> b n", b=2)[1])
-            instrumentals.append(instrumentals_ids)
-        vocals = np.concatenate(vocals, axis=1)
-        instrumentals = np.concatenate(instrumentals, axis=1)
-        
-        vocal_save_path = os.path.join(stage1_output_dir, f"vocal_{random_id}".replace('.', '@') + '.npy')
-        inst_save_path = os.path.join(stage1_output_dir, f"instrumental_{random_id}".replace('.', '@') + '.npy')
-        np.save(vocal_save_path, vocals)
-        np.save(inst_save_path, instrumentals)
-        stage1_output_set.append(vocal_save_path)
-        stage1_output_set.append(inst_save_path)
-        
+vocal_decoder.to(device).eval()
+inst_decoder.to(device).eval()
 
-        print("Converting to Audio...")
-
-        # convert audio tokens to audio
-        def save_audio(wav: torch.Tensor, path, sample_rate: int, rescale: bool = False):
-            folder_path = os.path.dirname(path)
-            if not os.path.exists(folder_path):
-                os.makedirs(folder_path)
-            limit = 0.99
-            max_val = wav.abs().max()
-            wav = wav * min(limit / max_val, 1) if rescale else wav.clamp(-limit, limit)
-            torchaudio.save(str(path), wav, sample_rate=sample_rate, encoding='PCM_S', bits_per_sample=16)
-
-        # reconstruct tracks
-        recons_output_dir = os.path.join(output_dir, "recons")
-        recons_mix_dir = os.path.join(recons_output_dir, 'mix')
-        os.makedirs(recons_mix_dir, exist_ok=True)
-        tracks = []
-        for npy in stage1_output_set:
-            codec_result = np.load(npy)
-            decodec_rlt = []
-            with torch.no_grad():
-                decoded_waveform = codec_model.decode(
-                    torch.as_tensor(codec_result.astype(np.int16), dtype=torch.long).unsqueeze(0).permute(1, 0, 2).to(
-                        device))
-            decoded_waveform = decoded_waveform.cpu().squeeze(0)
-            decodec_rlt.append(torch.as_tensor(decoded_waveform))
-            decodec_rlt = torch.cat(decodec_rlt, dim=-1)
-            save_path = os.path.join(recons_output_dir, os.path.splitext(os.path.basename(npy))[0] + ".mp3")
-            tracks.append(save_path)
-            save_audio(decodec_rlt, save_path, 16000)
-        # mix tracks
-        for inst_path in tracks:
-            try:
-                if (inst_path.endswith('.wav') or inst_path.endswith('.mp3')) \
-                        and 'instrumental' in inst_path:
-                    # find pair
-                    vocal_path = inst_path.replace('instrumental', 'vocal')
-                    if not os.path.exists(vocal_path):
-                        continue
-                    # mix
-                    recons_mix = os.path.join(recons_mix_dir,
-                                              os.path.basename(inst_path).replace('instrumental', 'mixed'))
-                    vocal_stem, sr = sf.read(inst_path)
-                    instrumental_stem, _ = sf.read(vocal_path)
-                    mix_stem = (vocal_stem + instrumental_stem) / 1
-                    sf.write(recons_mix, mix_stem, sr)
-            except Exception as e:
-                print(e)
-
-        # vocoder to upsample audios
-        vocoder_output_dir = os.path.join(output_dir, 'vocoder')
-        vocoder_stems_dir = os.path.join(vocoder_output_dir, 'stems')
-        vocoder_mix_dir = os.path.join(vocoder_output_dir, 'mix')
-        os.makedirs(vocoder_mix_dir, exist_ok=True)
-        os.makedirs(vocoder_stems_dir, exist_ok=True)
-        instrumental_output = None
-        vocal_output = None
-        for npy in stage1_output_set:
-            if 'instrumental' in npy:
-                # Process instrumental
-                instrumental_output = process_audio(
-                    npy,
-                    os.path.join(vocoder_stems_dir, 'instrumental.mp3'),
-                    rescale,
-                    argparse.Namespace(**locals()),  # Convert local variables to argparse.Namespace
-                    inst_decoder,
-                    codec_model
-                )
-            else:
-                # Process vocal
-                vocal_output = process_audio(
-                    npy,
-                    os.path.join(vocoder_stems_dir, 'vocal.mp3'),
-                    rescale,
-                    argparse.Namespace(**locals()),  # Convert local variables to argparse.Namespace
-                    vocal_decoder,
-                    codec_model
-                )
-        # mix tracks
-        try:
-            mix_output = instrumental_output + vocal_output
-            vocoder_mix = os.path.join(vocoder_mix_dir, os.path.basename(recons_mix))
-            save_audio(mix_output, vocoder_mix, 44100, rescale)
-            print(f"Created mix: {vocoder_mix}")
-        except RuntimeError as e:
-            print(e)
-            print(f"mix {vocoder_mix} failed! inst: {instrumental_output.shape}, vocal: {vocal_output.shape}")
-
-        # Post process
-        final_output_path = os.path.join(output_dir, os.path.basename(recons_mix))
-        replace_low_freq_with_energy_matched(
-            a_file=recons_mix,  # 16kHz
-            b_file=vocoder_mix,  # 48kHz
-            c_file=final_output_path,
-            cutoff_freq=5500.0
-        )
-        print("All process Done")
-        
-        # Load the final audio file and return the numpy array
-        final_audio, sr = torchaudio.load(final_output_path)
-        return (sr, final_audio.squeeze().numpy())
+# Tokenizer and codec tool
+mmtokenizer = _MMSentencePieceTokenizer("./mm_tokenizer_v0.2_hf/tokenizer.model")
+codectool = CodecManipulator("xcodec", 0, 1)
 
+def generate_music(genre_txt, lyrics_txt, max_new_tokens=5, run_n_segments=2, use_audio_prompt=False, audio_prompt_path="", prompt_start_time=0.0, prompt_end_time=30.0, rescale=False):
+    if use_audio_prompt and not audio_prompt_path:
+        raise FileNotFoundError("Please provide an audio prompt filepath when enabling 'use_audio_prompt'!")
+
+    max_new_tokens *= 100
+    top_p = 0.93
+    temperature = 1.0
+    repetition_penalty = 1.2
+
+    # Split lyrics into segments
+    def split_lyrics(lyrics):
+        pattern = r"\[(\w+)\](.*?)\n(?=\[|\Z)"
+        segments = re.findall(pattern, lyrics, re.DOTALL)
+        return [f"[{seg[0]}]\n{seg[1].strip()}\n\n" for seg in segments]
+
+    lyrics = split_lyrics(lyrics_txt + "\n")
+    full_lyrics = "\n".join(lyrics)
+    prompt_texts = [f"Generate music from the given lyrics segment by segment.\n[Genre] {genre_txt.strip()}\n{full_lyrics}"] + lyrics
+
+    raw_output = None
+    stage1_output_set = []
+
+    for i, p in enumerate(tqdm(prompt_texts[:run_n_segments])):
+        section_text = p.replace('[start_of_segment]', '').replace('[end_of_segment]', '')
+        guidance_scale = 1.5 if i <= 1 else 1.2
+
+        if i == 0:
+            continue
+
+        if i == 1 and use_audio_prompt:
+            audio_prompt = load_audio_mono(audio_prompt_path)
+            audio_prompt = audio_prompt.unsqueeze(0).to(device)
+            raw_codes = codec_model.encode(audio_prompt, target_bw=0.5).transpose(0, 1).cpu().numpy().astype(np.int16)
+            audio_prompt_codec = codectool.npy2ids(raw_codes[0])[int(prompt_start_time * 50): int(prompt_end_time * 50)]
+            audio_prompt_codec_ids = [mmtokenizer.soa] + codectool.sep_ids + audio_prompt_codec + [mmtokenizer.eoa]
+            sentence_ids = mmtokenizer.tokenize("[start_of_reference]") + audio_prompt_codec_ids + mmtokenizer.tokenize("[end_of_reference]")
+            head_id = mmtokenizer.tokenize(prompt_texts[0]) + sentence_ids
+        else:
+            head_id = mmtokenizer.tokenize(prompt_texts[0])
+
+        prompt_ids = head_id + mmtokenizer.tokenize(section_text) + [mmtokenizer.soa] + codectool.sep_ids
+        prompt_ids = torch.as_tensor(prompt_ids).unsqueeze(0).to(device)
+
+        input_ids = torch.cat([raw_output, prompt_ids], dim=1) if i > 1 else prompt_ids
+
+        max_context = 16384 - max_new_tokens - 1
+        if input_ids.shape[-1] > max_context:
+            input_ids = input_ids[:, -(max_context):]
+
+        with torch.inference_mode(), torch.autocast(device_type='cuda', dtype=torch.float16):
+            output_seq = model.generate(
+                input_ids=input_ids,
+                max_new_tokens=max_new_tokens,
+                min_new_tokens=100,
+                do_sample=True,
+                top_p=top_p,
+                temperature=temperature,
+                repetition_penalty=repetition_penalty,
+                eos_token_id=mmtokenizer.eoa,
+                pad_token_id=mmtokenizer.eoa,
+                logits_processor=LogitsProcessorList([
+                    BlockTokenRangeProcessor(0, 32002),
+                    BlockTokenRangeProcessor(32016, 32016)
+                ]),
+                guidance_scale=guidance_scale,
+                use_cache=True,
+                top_k=50,
+                num_beams=1
+            )
+
+        if output_seq[0][-1].item() != mmtokenizer.eoa:
+            tensor_eoa = torch.as_tensor([[mmtokenizer.eoa]]).to(device)
+            output_seq = torch.cat((output_seq, tensor_eoa), dim=1)
+
+        raw_output = torch.cat([raw_output, prompt_ids, output_seq[:, input_ids.shape[-1]:]], dim=1) if i > 1 else output_seq
+
+    # Process and save outputs
+    ids = raw_output[0].cpu().numpy()
+    soa_idx = np.where(ids == mmtokenizer.soa)[0].tolist()
+    eoa_idx = np.where(ids == mmtokenizer.eoa)[0].tolist()
+
+    vocals, instrumentals = [], []
+    for i in range(len(soa_idx)):
+        codec_ids = ids[soa_idx[i] + 1:eoa_idx[i]]
+        if codec_ids[0] == 32016:
+            codec_ids = codec_ids[1:]
+        codec_ids = codec_ids[:2 * (codec_ids.shape[0] // 2)]
+        vocals.append(codectool.ids2npy(rearrange(codec_ids, "(n b) -> b n", b=2)[0]))
+        instrumentals.append(codectool.ids2npy(rearrange(codec_ids, "(n b) -> b n", b=2)[1]))
+
+    vocals = np.concatenate(vocals, axis=1)
+    instrumentals = np.concatenate(instrumentals, axis=1)
+
+    # Decode and mix audio
+    decoded_vocals = codec_model.decode(torch.as_tensor(vocals.astype(np.int16), dtype=torch.long).unsqueeze(0).permute(1, 0, 2).to(device)).cpu().squeeze(0)
+    decoded_instrumentals = codec_model.decode(torch.as_tensor(instrumentals.astype(np.int16), dtype=torch.long).unsqueeze(0).permute(1, 0, 2).to(device)).cpu().squeeze(0)
+
+    mixed_audio = (decoded_vocals + decoded_instrumentals) / 2
+    return (16000, mixed_audio.numpy())
 
 @spaces.GPU(duration=120)
 def infer(genre_txt_content, lyrics_txt_content, num_segments=2, max_new_tokens=10):
-    # Execute the command
     try:
-        audio_data = generate_music(genre_txt=genre_txt_content, lyrics_txt=lyrics_txt_content, run_n_segments=num_segments,
-                               cuda_idx=0, max_new_tokens=max_new_tokens)
-        return audio_data
+        return generate_music(genre_txt=genre_txt_content, lyrics_txt=lyrics_txt_content, run_n_segments=num_segments, max_new_tokens=max_new_tokens)
     except Exception as e:
-        gr.Warning("An Error Occured: " + str(e))
+        gr.Warning("An Error Occurred: " + str(e))
         return None
-    finally:
-        print("Temporary files deleted.")
-
 
-# Gradio
+# Gradio Interface
 with gr.Blocks() as demo:
     with gr.Column():
         gr.Markdown("# YuE: Open Music Foundation Models for Full-Song Generation")
@@ -493,4 +312,5 @@ Living out my dreams with this mic and a deal
         inputs=[genre_txt, lyrics_txt, num_segments, max_new_tokens],
         outputs=[music_out]
     )
+
 demo.queue().launch(show_error=True)