File size: 4,620 Bytes
71e354e | 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 | /**************************************************************************************************
*
* Copyright (c) 2019-2026 Axera Semiconductor (Ningbo) Co., Ltd. All Rights Reserved.
*
* This source file is the property of Axera Semiconductor (Ningbo) Co., Ltd. and
* may not be copied or distributed in any isomorphic form without the prior
* written consent of Axera Semiconductor (Ningbo) Co., Ltd.
*
**************************************************************************************************/
#include <stdio.h>
#include <unistd.h>
#include <limits.h>
#include <cstdlib>
#include <cstring>
#include "utils/cmdline.hpp"
#include "utils/logger.h"
#include "utils/AudioFile.h"
#include "api/ax_tts_api.h"
static std::string pick_path(const char* env_var, const char* default_rel) {
const char* env = std::getenv(env_var);
if (env && env[0] != '\0') {
return std::string(env);
}
return std::string(default_rel);
}
int main(int argc, char** argv) {
cmdline::parser cmd;
cmd.add<std::string>("language", 'l', "Language, in ISO-639 format", true, "en");
cmd.add<std::string>("text", 't', "Input text", true, "");
cmd.add<std::string>("output", 'o', "Output wav path", true, "");
cmd.add<std::string>("model_path", 0, "Model path (override default)", false, "");
cmd.add<std::string>("espeak_data_path", 0, "Path to espeak-ng-data directory", false, "");
cmd.add<std::string>("jieba_dict_path", 0, "Path to jieba dict directory", false, "");
cmd.parse_check(argc, argv);
// 0. get app args, can be removed from user's app
auto input_text = cmd.get<std::string>("text");
auto language = cmd.get<std::string>("language");
auto output = cmd.get<std::string>("output");
std::string voice;
if (language == "ja") voice = "jf_gongitsune";
else if (language == "zh") voice = "zf_xiaoxiao";
else voice = "af_heart";
AX_TTS_INIT_CONFIG init_config;
init_config.max_seq_len = 96;
#if defined(CHIP_AX650) || defined(CHIP_AX8850)
std::string model_path = "models-ax650";
#else
std::string model_path = "models-ax630c";
#endif
if (!cmd.get<std::string>("model_path").empty()) {
model_path = cmd.get<std::string>("model_path");
}
if (model_path.size() >= AX_TTS_MAX_STR_LEN) {
ALOGE("model_path too long for AX_TTS_MAX_STR_LEN");
return -1;
}
snprintf(init_config.model_path, AX_TTS_MAX_STR_LEN, "%s", model_path.c_str());
std::string espeak_path = cmd.get<std::string>("espeak_data_path");
if (espeak_path.empty()) espeak_path = pick_path("AX_TTS_ESPEAK_DATA_PATH", "espeak-ng-data");
std::string jieba_path = cmd.get<std::string>("jieba_dict_path");
if (jieba_path.empty()) jieba_path = pick_path("AX_TTS_JIEBA_DICT_PATH", "dict");
if (espeak_path.size() >= AX_TTS_MAX_STR_LEN || jieba_path.size() >= AX_TTS_MAX_STR_LEN) {
ALOGE("espeak/jieba path too long for AX_TTS_MAX_STR_LEN");
return -1;
}
snprintf(init_config.espeak_data_path, AX_TTS_MAX_STR_LEN, "%s", espeak_path.c_str());
snprintf(init_config.jieba_dict_path, AX_TTS_MAX_STR_LEN, "%s", jieba_path.c_str());
AX_TTS_HANDLE handle = AX_TTS_Init(AX_KOKORO, &init_config);
if (!handle) {
ALOGE("AX_TTS_Init failed!");
return -1;
}
AX_TTS_RUN_CONFIG run_config;
run_config.fade_out = 0.3f;
run_config.speed = 1.0f;
run_config.sample_rate = 24000;
snprintf(run_config.language, AX_TTS_MAX_STR_LEN, "%s", language.c_str());
snprintf(run_config.voice, AX_TTS_MAX_STR_LEN, "%s", voice.c_str());
AX_TTS_AUDIO* audio = NULL;
int ret = AX_TTS_Run(handle,
input_text.c_str(),
&run_config,
&audio);
if (ret != 0) {
ALOGE("AX_TTS_Run failed!");
free(audio);
return -1;
}
AudioFile<float> audio_file;
std::vector<std::vector<float> > audio_samples{std::vector<float>(audio->data, audio->data + audio->num_samples)};
audio_file.setAudioBuffer(audio_samples);
audio_file.setSampleRate(run_config.sample_rate);
if (!audio_file.save(output)) {
ALOGE("Save audio file failed!\n");
free(audio);
return -1;
}
free(audio);
printf("================================\n");
printf("test_zh:\n");
printf("input text: %s\n", input_text.c_str());
printf("output duration: %.2f seconds\n", audio_file.getNumSamplesPerChannel() * 1.0f / run_config.sample_rate);
printf("output file: %s\n", output.c_str());
printf("\n");
return 0;
}
|