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
99 lines
3.5 KiB
C++
99 lines
3.5 KiB
C++
#pragma once
|
|
|
|
#include "engine/models/vevo2/assets.h"
|
|
#include "engine/models/vevo2/types.h"
|
|
|
|
#include <cstddef>
|
|
#include <cstdint>
|
|
#include <memory>
|
|
#include <vector>
|
|
|
|
namespace engine::assets {
|
|
enum class TensorStorageType;
|
|
}
|
|
|
|
namespace engine::core {
|
|
class ExecutionContext;
|
|
}
|
|
|
|
namespace engine::models::vevo2 {
|
|
|
|
struct Vevo2CocoTokenizerWeights;
|
|
struct Vevo2CocoTokenizerGraph;
|
|
|
|
class Vevo2CocoProsodyTokenizerRuntime final {
|
|
public:
|
|
Vevo2CocoProsodyTokenizerRuntime(
|
|
const Vevo2Assets & assets,
|
|
engine::core::ExecutionContext & execution_context,
|
|
size_t weight_context_bytes,
|
|
size_t graph_context_bytes,
|
|
engine::assets::TensorStorageType matmul_weight_storage_type,
|
|
engine::assets::TensorStorageType conv_weight_storage_type);
|
|
~Vevo2CocoProsodyTokenizerRuntime();
|
|
|
|
Vevo2TokenSequence encode(
|
|
const runtime::AudioBuffer & prosody_audio,
|
|
const std::optional<runtime::AudioBuffer> & style_ref_audio,
|
|
const Vevo2GenerationOptions & generation) const;
|
|
Vevo2TokenSequence encode_chromagram_features(
|
|
const std::vector<float> & chromagram_features,
|
|
int64_t feature_frames) const;
|
|
|
|
private:
|
|
Vevo2CocoTokenizerGraph & ensure_graph(int64_t feature_frames, bool uses_whisper) const;
|
|
|
|
Vevo2CocoTokenizerConfig config_;
|
|
engine::core::ExecutionContext & execution_context_;
|
|
size_t graph_context_bytes_ = 0;
|
|
std::shared_ptr<const engine::assets::TensorSource> weight_source_;
|
|
std::shared_ptr<const Vevo2CocoTokenizerWeights> weights_;
|
|
mutable std::unique_ptr<Vevo2CocoTokenizerGraph> graph_;
|
|
std::string name_;
|
|
};
|
|
|
|
class Vevo2CocoContentStyleTokenizerRuntime final {
|
|
public:
|
|
Vevo2CocoContentStyleTokenizerRuntime(
|
|
const Vevo2Assets & assets,
|
|
engine::core::ExecutionContext & execution_context,
|
|
size_t weight_context_bytes,
|
|
size_t graph_context_bytes,
|
|
engine::assets::TensorStorageType matmul_weight_storage_type,
|
|
engine::assets::TensorStorageType conv_weight_storage_type);
|
|
~Vevo2CocoContentStyleTokenizerRuntime();
|
|
|
|
Vevo2TokenSequence encode_style_reference(
|
|
const std::optional<runtime::AudioBuffer> & style_ref_audio,
|
|
const std::optional<std::vector<float>> & whisper_features,
|
|
int64_t feature_frames,
|
|
const Vevo2GenerationOptions & generation) const;
|
|
Vevo2TokenSequence encode_timbre_reference(
|
|
const runtime::AudioBuffer & timbre_ref_audio,
|
|
const std::vector<float> & whisper_features,
|
|
int64_t feature_frames) const;
|
|
Vevo2TokenSequence encode_feature_frames(
|
|
const std::vector<float> & whisper_features,
|
|
const std::vector<float> & chromagram_features,
|
|
int64_t feature_frames) const;
|
|
Vevo2TokenSequence encode_shifted_reference(
|
|
const runtime::AudioBuffer & audio,
|
|
const std::vector<float> & whisper_features,
|
|
int64_t feature_frames,
|
|
int pitch_shift_steps) const;
|
|
|
|
private:
|
|
Vevo2CocoTokenizerGraph & ensure_graph(int64_t feature_frames, bool uses_whisper) const;
|
|
|
|
Vevo2CocoTokenizerConfig config_;
|
|
engine::core::ExecutionContext & execution_context_;
|
|
size_t graph_context_bytes_ = 0;
|
|
std::shared_ptr<const engine::assets::TensorSource> weight_source_;
|
|
std::shared_ptr<const Vevo2CocoTokenizerWeights> weights_;
|
|
mutable std::unique_ptr<Vevo2CocoTokenizerGraph> graph_;
|
|
std::vector<float> whisper_mean_;
|
|
std::vector<float> whisper_std_;
|
|
std::string name_;
|
|
};
|
|
|
|
} // namespace engine::models::vevo2
|