vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-07-09 17:08:33 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fixed tts mp3 saving

Browse source
This commit is contained in:
Concedo 2026-06-18 22:33:30 +08:00
parent 1b36e7f606
commit 45f49f9bd9
3 changed files with 62 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

6
otherarch/tts_adapter.cpp
View file

@ -498,11 +498,7 @@ static std::string save_tts_audio_base64(const std::vector<float> & audio, int s
return save_wav16_base64(audio, sample_rate);
}
std::vector<float> stereo_audio;
stereo_audio.reserve(audio.size() * 2);
stereo_audio.insert(stereo_audio.end(), audio.begin(), audio.end());
stereo_audio.insert(stereo_audio.end(), audio.begin(), audio.end());
return save_stereo_mp3_base64(stereo_audio, (int) audio.size(), sample_rate);
return save_mono_mp3_base64(audio, sample_rate);
}
//ttscpp specific

60
otherarch/utils.cpp
View file

@ -627,6 +627,66 @@ std::string save_stereo_wav16_base64(const std::vector<float> & raw_audio, int T
return kcpp_base64_encode(wav_data);
}
std::string save_mono_mp3_base64(const std::vector<float> & raw_audio, int sample_rate) {
std::vector<float> limited_audio = raw_audio;
float peak = 0.0f;
for (float s : limited_audio) {
peak = std::max(peak, std::fabs(s));
}
if (peak > 1e-6f) {
const float target_amp = 0.89125094f; // -1.0 dBFS
const float gain = peak > target_amp ? target_amp / peak : 1.0f;
for (float & s : limited_audio) {
s = std::max(-target_amp, std::min(target_amp, s * gain));
}
}
const float * enc_audio = limited_audio.data();
int enc_T = (int) limited_audio.size();
int enc_sr = sample_rate;
std::vector<float> resampled;
if (sample_rate != 32000 && sample_rate != 44100 && sample_rate != 48000) {
resampled = resample_wav(1, limited_audio, sample_rate, 44100);
enc_audio = resampled.data();
enc_sr = 44100;
enc_T = (int) resampled.size();
}
if (enc_T <= 0) {
return "";
}
mp3enc_t * enc = mp3enc_init(enc_sr, 1, 128);
if (!enc) {
fprintf(stderr, "[Audio] mp3enc_init failed\n");
return "";
}
std::string mp3_data;
mp3_data.reserve(enc_T / 4);
int chunk_size = enc_sr;
std::vector<float> buf((size_t) chunk_size);
for (int pos = 0; pos < enc_T; pos += chunk_size) {
int n = (pos + chunk_size <= enc_T) ? chunk_size : (enc_T - pos);
memcpy(buf.data(), enc_audio + pos, (size_t) n * sizeof(float));
int out_size = 0;
const uint8_t * mp3 = mp3enc_encode(enc, buf.data(), n, &out_size);
if (out_size > 0) {
mp3_data.append((const char *) mp3, out_size);
}
}
int flush_size = 0;
const uint8_t * flush_data = mp3enc_flush(enc, &flush_size);
if (flush_size > 0) {
mp3_data.append((const char *) flush_data, flush_size);
}
mp3enc_free(enc);
return kcpp_base64_encode(mp3_data);
}
std::string save_stereo_mp3_base64(const std::vector<float> & raw_audio,int T_audio,int sample_rate) {
const float * enc_audio = raw_audio.data();
int enc_T = T_audio;

1
otherarch/utils.h
View file

@ -114,3 +114,4 @@ std::string save_ulaw_wav8_base64(const std::vector<float> &data, int sample_rat
std::string save_wav16_base64(const std::vector<float> &data, int sample_rate);
std::string save_stereo_wav16_base64(const std::vector<float> & raw_audio, int T_audio, int sample_rate);
std::string save_stereo_mp3_base64(const std::vector<float> & raw_audio,int T_audio,int sample_rate);
std::string save_mono_mp3_base64(const std::vector<float> & raw_audio, int sample_rate);