File size: 7,010 Bytes
5fe16b1
15d0727
5fe16b1
 
 
 
 
 
 
 
 
 
187ec28
4d25f3d
5fe16b1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
187ec28
 
 
5fe16b1
 
187ec28
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5fe16b1
 
 
 
187ec28
 
 
 
 
 
 
 
 
5fe16b1
 
 
 
 
 
885ea0a
5fe16b1
 
 
 
 
 
 
 
 
 
 
187ec28
 
 
 
 
 
 
 
5fe16b1
41484d1
5fe16b1
 
 
187ec28
5fe16b1
187ec28
5fe16b1
187ec28
885ea0a
5fe16b1
41484d1
 
 
 
187ec28
 
885ea0a
41484d1
bf07215
 
 
4d25f3d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
from dataclasses import dataclass
from typing import List, Tuple, Dict
import os
import re
import httpx
import json
from openai import OpenAI
import edge_tts
import tempfile
from pydub import AudioSegment
import base64
from pathlib import Path
import hashlib
import asyncio

@dataclass
class ConversationConfig:
    max_words: int = 3000
    prefix_url: str = "https://r.jina.ai/"
    model_name: str = "meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo"

class URLToAudioConverter:
    def __init__(self, config: ConversationConfig, llm_api_key: str):
        self.config = config
        self.llm_client = OpenAI(api_key=llm_api_key, base_url="https://api.together.xyz/v1")
        self.llm_out = None

    def fetch_text(self, url: str) -> str:
        if not url:
            raise ValueError("URL cannot be empty")
        response = httpx.get(f"{self.config.prefix_url}{url}", timeout=60.0)
        response.raise_for_status()
        return response.text

    def extract_conversation(self, text: str) -> Dict:
        prompt = (
            f"{text}\nConvert the provided text into a short informative podcast conversation "
            f"between two experts. Return ONLY a JSON object with the following structure:\n"
            '{"conversation": [{"speaker": "Speaker1", "text": "..."}, {"speaker": "Speaker2", "text": "..."}]}'
        )
        chat_completion = self.llm_client.chat.completions.create(
            messages=[{"role": "user", "content": prompt}],
            model=self.config.model_name,
            response_format={"type": "json_object"}
        )
        response_content = chat_completion.choices[0].message.content
        json_str = response_content.strip()
        if not json_str.startswith("{"):
            json_str = json_str[json_str.find("{"):]
        if not json_str.endswith("}"):
            json_str = json_str[: json_str.rfind("}") + 1]
        return json.loads(json_str)

    async def text_to_speech(self, conversation_json: Dict, voice_1: str, voice_2: str) -> Tuple[List[str], str]:
        output_dir = Path(self._create_output_directory())
        filenames = []
        for i, turn in enumerate(conversation_json["conversation"]):
            voice = voice_1 if i % 2 == 0 else voice_2
            tmp_path, error = await self._generate_audio(turn["text"], voice)
            if error:
                raise RuntimeError(f"Text-to-speech failed: {error}")
            filename = output_dir / f"output_{i}.mp3"
            os.rename(tmp_path, filename)
            filenames.append(str(filename))
        return filenames, str(output_dir)

    async def _generate_audio(self, text: str, voice: str, rate: int = 0, pitch: int = 0) -> Tuple[str, str]:
        voice_short_name = voice.split(" - ")[0]
        rate_str = f"{rate:+d}%"
        pitch_str = f"{pitch:+d}Hz"
        communicate = edge_tts.Communicate(text, voice_short_name, rate=rate_str, pitch=pitch_str)
        with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp_file:
            tmp_path = tmp_file.name
            await communicate.save(tmp_path)
        return tmp_path, None

    def _create_output_directory(self) -> str:
        random_bytes = os.urandom(8)
        folder_name = base64.urlsafe_b64encode(random_bytes).decode("utf-8")
        os.makedirs(folder_name, exist_ok=True)
        return folder_name

    def combine_audio_files(self, filenames: List[str], output_file: str) -> None:
        combined = AudioSegment.empty()
        for filename in filenames:
            combined += AudioSegment.from_file(filename, format="mp3")
        combined.export(output_file, format="mp3")
        dir_path = os.path.dirname(filenames[0])
        for file in os.listdir(dir_path):
            os.remove(os.path.join(dir_path, file))
        os.rmdir(dir_path)

    async def url_to_audio(self, url: str, voice_1: str, voice_2: str) -> Tuple[str, str]:
        text = self.fetch_text(url)
        words = text.split()
        if len(words) > self.config.max_words:
            text = " ".join(words[: self.config.max_words])
        conversation_json = self.extract_conversation(text)
        conversation_text = "\n".join(f"{t['speaker']}: {t['text']}" for t in conversation_json["conversation"])
        self.llm_out = conversation_json
        audio_files, folder_name = await self.text_to_speech(conversation_json, voice_1, voice_2)
        final_output = os.path.join(folder_name, "combined_output.mp3")
        self.combine_audio_files(audio_files, final_output)
        return final_output, conversation_text

    async def text_to_audio(self, text: str, voice_1: str, voice_2: str) -> Tuple[str, str]:
        conversation_json = self.extract_conversation(text)
        conversation_text = "\n".join(f"{t['speaker']}: {t['text']}" for t in conversation_json["conversation"])
        audio_files, folder_name = await self.text_to_speech(conversation_json, voice_1, voice_2)
        final_output = os.path.join(folder_name, "combined_output.mp3")
        self.combine_audio_files(audio_files, final_output)
        return final_output, conversation_text

    async def raw_text_to_audio(self, text: str, voice_1: str, voice_2: str) -> Tuple[str, str]:
        try:
            print("\n=== DEBUG INICIO (raw_text_to_audio) ===")
            print(f"Texto recibido: {text[:200]}...")  # Verifica el input
            
            # Usa una ruta absoluta en /tmp (compatible con Spaces)
            output_dir = "/tmp/podcast_outputs"
            os.makedirs(output_dir, exist_ok=True)
            hash_name = hashlib.md5(text.encode()).hexdigest()[:8]
            output_file = os.path.join(output_dir, f"podcast_{hash_name}.mp3")
            print(f"Ruta de salida: {output_file}")

            # Verifica voces disponibles (DEBUG)
            voices = await edge_tts.list_voices()
            voice_names = [v['Name'] for v in voices]
            print(f"Voces disponibles (primeras 5): {voice_names[:5]}...")

            # Extrae el nombre corto de la voz (ej: "en-US-AvaMultilingualNeural")
            voice_short = voice_1.split(" - ")[0] if " - " in voice_1 else voice_1
            print(f"Voz a usar: {voice_short}")

            # Genera el audio
            communicate = edge_tts.Communicate(text, voice_short)
            print("Generando audio...")
            await communicate.save(output_file)
            print("Audio generado.")

            # Verifica que el archivo existe y no está vacío
            if not os.path.exists(output_file):
                print("ERROR: Archivo no creado.")
                return "Error: Archivo no generado", None
            elif os.path.getsize(output_file) == 0:
                print("ERROR: Archivo vacío.")
                return "Error: Archivo de audio vacío", None

            print(f"=== DEBUG FIN (Archivo válido: {output_file}) ===")
            return text, output_file

        except Exception as e:
            print(f"ERROR CRÍTICO: {str(e)}")
            return f"Error: {str(e)}", None