voice_audio_multi/include/engine/models/ming3_audio/frontend.h
Fabio c36510c21e
Some checks failed
Linux Build / Linux cpu (push) Has been cancelled
Linux Build / Linux cpu-clang (push) Has been cancelled
Linux Build / Linux vulkan (push) Has been cancelled
mac-build / macOS CPU (push) Has been cancelled
Nix Build / Nix (cpu) (push) Has been cancelled
Nix Build / Nix (vulkan) (push) Has been cancelled
windows-build / Windows CPU (push) Has been cancelled
Initial import: audio.cpp with Orange Pi documentation
2026-11-04 19:05:22 +01:00

30 lines
937 B
C++

#pragma once
#include "engine/models/ming3_audio/assets.h"
#include "engine/framework/audio/dsp.h"
#include "engine/framework/runtime/session.h"
#include "engine/framework/tokenizers/llama_bpe.h"
namespace engine::models::ming3_audio {
struct AudioFeatures {
std::vector<float> log_mel;
int64_t frames = 0;
};
class Frontend {
public:
explicit Frontend(std::shared_ptr<const MingAudioAssets> assets);
AudioFeatures extract(const runtime::AudioBuffer & audio, size_t threads) const;
std::vector<int32_t> prompt(const std::string & instruction, int64_t mel_frames = 0,
const std::string & system_prompt = {}, bool enable_thinking = false) const;
std::string decode(const std::vector<int32_t> & ids) const;
private:
std::shared_ptr<const MingAudioAssets> assets_;
tokenizers::LlamaBpeTokenizer tokenizer_;
audio::WhisperLogMelExtractor mel_;
};
} // namespace engine::models::ming3_audio