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
69 lines
2.6 KiB
C++
69 lines
2.6 KiB
C++
#pragma once
|
|
|
|
#include "engine/framework/assets/tensor_source.h"
|
|
#include "engine/framework/runtime/session_base.h"
|
|
#include "engine/models/supertonic/assets.h"
|
|
#include "engine/models/supertonic/tokenizer_text.h"
|
|
|
|
#include <cstddef>
|
|
#include <memory>
|
|
#include <optional>
|
|
#include <string>
|
|
#include <vector>
|
|
|
|
namespace engine::models::supertonic {
|
|
|
|
class SupertonicRuntime;
|
|
|
|
struct SupertonicGenerationOptions {
|
|
int num_inference_steps = 8;
|
|
float speaking_rate = 1.05F;
|
|
uint32_t seed = 1234U;
|
|
std::string voice = "M1";
|
|
std::string language = "en";
|
|
};
|
|
|
|
class SupertonicSession final
|
|
: public runtime::RuntimeSessionBase
|
|
, public runtime::IOfflineVoiceTaskSession
|
|
, public runtime::IStreamingVoiceTaskSession {
|
|
public:
|
|
SupertonicSession(
|
|
runtime::TaskSpec task,
|
|
runtime::SessionOptions options,
|
|
std::shared_ptr<const SupertonicAssets> assets);
|
|
~SupertonicSession() override;
|
|
|
|
std::string family() const override;
|
|
runtime::VoiceTaskKind task_kind() const override;
|
|
runtime::RunMode run_mode() const override;
|
|
void prepare(const runtime::SessionPreparationRequest & request) override;
|
|
runtime::TaskResult run(const runtime::TaskRequest & request) override;
|
|
runtime::StreamingPolicy streaming_policy() const override;
|
|
void start_stream(const runtime::TaskRequest & request) override;
|
|
std::optional<runtime::StreamEvent> next_stream_event() override;
|
|
void set_stream_event_sink(runtime::StreamEventCallback sink) override;
|
|
runtime::TaskResult finish_stream() override;
|
|
void reset() override;
|
|
runtime::StreamEvent process_audio_chunk(const runtime::AudioChunk & chunk) override;
|
|
runtime::TaskResult finalize() override;
|
|
|
|
private:
|
|
SupertonicGenerationOptions generation_options_from_request(const runtime::TaskRequest & request) const;
|
|
void validate_request(const runtime::TaskRequest & request) const;
|
|
std::vector<runtime::TaskRequest> build_chunk_requests(const runtime::TaskRequest & request) const;
|
|
runtime::AudioBuffer synthesize_chunk(const runtime::TaskRequest & request);
|
|
|
|
runtime::TaskSpec task_;
|
|
std::shared_ptr<const SupertonicAssets> assets_;
|
|
SupertonicTextTokenizer tokenizer_;
|
|
assets::TensorStorageType weight_storage_type_ = assets::TensorStorageType::Native;
|
|
std::size_t style_cache_slots_ = 4;
|
|
std::unique_ptr<SupertonicRuntime> runtime_;
|
|
std::vector<runtime::TaskRequest> stream_chunk_requests_;
|
|
runtime::AudioBuffer stream_merged_audio_;
|
|
std::size_t stream_chunk_index_ = 0;
|
|
bool stream_started_ = false;
|
|
};
|
|
|
|
} // namespace engine::models::supertonic
|