diff --git a/examples/example_non_streaming.py b/examples/example_non_streaming.py new file mode 100644 index 0000000..bf5e41b --- /dev/null +++ b/examples/example_non_streaming.py @@ -0,0 +1,206 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- +""" +FlowTTS Non-Streaming Example - TextToSpeech with MP3/PCM support + +This example demonstrates the non-streaming TTS API which returns +the complete audio in a single response. Supports MP3 and PCM formats. + +API Reference: +- https://cloud.tencent.com/document/api/647/122475 +- https://cloud.tencent.com/document/api/647/44055#AudioFormat +""" + +import os +import io +import json +import time +import wave +import base64 +from dotenv import load_dotenv +from tencentcloud.common import credential +from tencentcloud.common.profile.client_profile import ClientProfile +from tencentcloud.common.profile.http_profile import HttpProfile +from tencentcloud.trtc.v20190722 import trtc_client, models + +# Load environment variables from .env file +load_dotenv() + +# ========== Configuration ========== +SDK_APP_ID = int(os.getenv("TENCENTCLOUD_SDK_APP_ID", "1400000000")) +MODEL = "flow_01_turbo" + +VOICE_CONFIG = { + "VoiceId": "v-female-R2s4N9qJ", + "Speed": 1.0, + "Volume": 1.0, + "Pitch": 0, + "Language": "zh" +} + +# Audio format configuration +# For non-streaming API, both "mp3" and "pcm" are supported +AUDIO_FORMAT = { + "Format": "mp3", # "mp3" or "pcm" + "SampleRate": 24000 # 16000 or 24000 +} + +TEST_TEXT = "欢迎使用腾讯云语音合成服务，祝您使用愉快！" +# =================================== + + +def create_client(): + """Create Tencent Cloud TRTC client.""" + secret_id = os.getenv("TENCENTCLOUD_SECRET_ID") + secret_key = os.getenv("TENCENTCLOUD_SECRET_KEY") + + if not secret_id or not secret_key: + print("Error: TENCENTCLOUD_SECRET_ID and TENCENTCLOUD_SECRET_KEY are required") + return None + + cred = credential.Credential(secret_id, secret_key) + + http_profile = HttpProfile() + http_profile.endpoint = os.getenv( + "TENCENTCLOUD_ENDPOINT", + "trtc.ai.tencentcloudapi.com" + ) + http_profile.reqTimeout = 120 + http_profile.keepAlive = True + http_profile.pre_conn_pool_size = 3 + + client_profile = ClientProfile() + client_profile.httpProfile = http_profile + + return trtc_client.TrtcClient(cred, "ap-beijing", client_profile) + + +def pcm_to_wav(pcm_data, sample_rate=24000, channels=1, sample_width=2): + """Convert PCM data to WAV format.""" + wav_buffer = io.BytesIO() + with wave.open(wav_buffer, 'wb') as wav_file: + wav_file.setnchannels(channels) + wav_file.setsampwidth(sample_width) + wav_file.setframerate(sample_rate) + wav_file.writeframes(pcm_data) + return wav_buffer.getvalue() + + +def save_audio_file(audio_data, filename): + """Save audio data to file.""" + output_dir = os.path.dirname(os.path.abspath(__file__)) + filepath = os.path.join(output_dir, filename) + + with open(filepath, 'wb') as f: + f.write(audio_data) + + print(f" Audio saved to: {filepath}") + return filepath + + +def text_to_speech_non_streaming(text=None, audio_format="mp3"): + """ + Non-streaming TTS API. + + Args: + text: Text to synthesize (default: TEST_TEXT) + audio_format: Output format, "mp3" or "pcm" (default: "mp3") + + Returns: + Path to the saved audio file, or None on failure + """ + if text is None: + text = TEST_TEXT + + print(f"\nStarting TTS synthesis") + print(f" Voice: {VOICE_CONFIG['VoiceId']}") + print(f" Text: {text}") + print(f" Text length: {len(text)}") + print(f" Format: {audio_format}") + + try: + client = create_client() + if not client: + return None + + req = models.TextToSpeechRequest() + + # Configure audio format + audio_config = { + "Format": audio_format, + "SampleRate": AUDIO_FORMAT["SampleRate"] + } + + params = { + "Model": MODEL, + "Text": text, + "Voice": VOICE_CONFIG, + "AudioFormat": audio_config, + "SdkAppId": SDK_APP_ID + } + req.from_json_string(json.dumps(params)) + + print(f" SdkAppId: {SDK_APP_ID}") + + start_time = time.time() + resp = client.TextToSpeech(req) + end_time = time.time() + + if hasattr(resp, 'Audio') and resp.Audio: + audio_b64 = resp.Audio + audio_data = base64.b64decode(audio_b64) + + # Print RequestId + request_id = getattr(resp, 'RequestId', 'N/A') + print(f" RequestId: {request_id}") + + # Handle audio based on format + if audio_format == "mp3": + # MP3 format - save directly + print(f" MP3 data size: {len(audio_data)} bytes") + print(f" Time elapsed: {(end_time - start_time) * 1000:.0f}ms") + filename = f"tts_{VOICE_CONFIG['VoiceId']}_{int(time.time())}.mp3" + result = save_audio_file(audio_data, filename) + else: + # PCM format - convert to WAV + wav_data = pcm_to_wav(audio_data, sample_rate=AUDIO_FORMAT["SampleRate"]) + print(f" WAV data size: {len(wav_data)} bytes") + print(f" Time elapsed: {(end_time - start_time) * 1000:.0f}ms") + filename = f"tts_{VOICE_CONFIG['VoiceId']}_{int(time.time())}.wav" + result = save_audio_file(wav_data, filename) + + if result: + print(f"\nTTS synthesis successful!") + return result + else: + print("Error: No audio data in response") + return None + + except Exception as e: + print(f"Error: {e}") + return None + + +def main(): + """Run examples for both MP3 and PCM formats.""" + # Example 1: MP3 format (default) + print("=" * 50) + print("Example 1: MP3 Format") + print("=" * 50) + text_to_speech_non_streaming( + "欢迎使用腾讯云FlowTTS，这是MP3格式输出。", + audio_format="mp3" + ) + + # Example 2: PCM format (converted to WAV) + print("\n" + "=" * 50) + print("Example 2: PCM Format (saved as WAV)") + print("=" * 50) + text_to_speech_non_streaming( + "欢迎使用腾讯云FlowTTS，这是PCM格式输出。", + audio_format="pcm" + ) + + +if __name__ == "__main__": + main()