108 lines
5.0 KiB
C
108 lines
5.0 KiB
C
|
/**
|
||
|
|
* Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
|
||
|
|
* SPDX-License-Identifier: Apache-2.0.
|
||
|
|
*/
|
||
|
|
|
||
|
|
#pragma once
|
||
|
|
|
||
|
|
#include <aws/text-to-speech/TextToSpeech_EXPORTS.h>
|
||
|
|
#include <aws/text-to-speech/PCMOutputDriver.h>
|
||
|
|
#include <aws/polly/PollyClient.h>
|
||
|
|
#include <aws/core/client/AsyncCallerContext.h>
|
||
|
|
#include <functional>
|
||
|
|
#include <memory>
|
||
|
|
#include <mutex>
|
||
|
|
|
||
|
|
namespace Aws
|
||
|
|
{
|
||
|
|
namespace TextToSpeech
|
||
|
|
{
|
||
|
|
/**
|
||
|
|
* Callback for handling notifications that the SendTextToOutputDevice operation has finished.
|
||
|
|
* @arg1 the text that was sent
|
||
|
|
* @arg2 the outcome of the operation from the polly service.
|
||
|
|
* @arg3 whether or not the audio stream was successfully sent to the audio driver.
|
||
|
|
*/
|
||
|
|
typedef std::function<void(const char*, const Polly::Model::SynthesizeSpeechOutcome&, bool)> SendTextCompletedHandler;
|
||
|
|
|
||
|
|
typedef std::pair<DeviceInfo, std::shared_ptr<PCMOutputDriver>> OutputDevicePair;
|
||
|
|
typedef Aws::Vector<OutputDevicePair> OutputDeviceList;
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Maximum bytes size of audio to be sent to PCM drivers by TextToSpeechManager in one operation
|
||
|
|
* with PCMOutputDriver::WriteBufferToDevice()
|
||
|
|
*/
|
||
|
|
static const size_t BUFF_SIZE = 8192;
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Manager for rendering text to the Polly service and then sending directly to an audio driver.
|
||
|
|
* By default this uses our best guess at the correct drivers for you operating system.
|
||
|
|
* On windows, this is the WaveOut API.
|
||
|
|
* On Linux, this is PulseAudio
|
||
|
|
* On Apple, this is CoreAudio.
|
||
|
|
*
|
||
|
|
* The drivers used can be arbitrarily overridden to send the stream anywhere you want. Simply provide your own driverFactory implementation.
|
||
|
|
*/
|
||
|
|
class AWS_TEXT_TO_SPEECH_API TextToSpeechManager : public std::enable_shared_from_this<TextToSpeechManager>
|
||
|
|
{
|
||
|
|
public:
|
||
|
|
/**
|
||
|
|
* Create a TextToSpeechManager instance initialized with a polly client and a driver factory.
|
||
|
|
* If driver factory is nullptr, we will create a default implementation for your operating system.
|
||
|
|
*/
|
||
|
|
static std::shared_ptr<TextToSpeechManager> Create(const std::shared_ptr<Polly::PollyClient>& pollyClient,
|
||
|
|
const std::shared_ptr<PCMOutputDriverFactory>& driverFactory = nullptr);
|
||
|
|
|
||
|
|
~TextToSpeechManager();
|
||
|
|
TextToSpeechManager(const TextToSpeechManager&) = delete;
|
||
|
|
TextToSpeechManager& operator=(const TextToSpeechManager&) = delete;
|
||
|
|
TextToSpeechManager(TextToSpeechManager&&) = delete;
|
||
|
|
TextToSpeechManager& operator=(TextToSpeechManager&&) = delete;
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Sends @text to the Polly Service, once the audio stream is returned, the audio stream is sent to your audio driver.
|
||
|
|
* @callback will be invoked once the entire operation has finished.
|
||
|
|
*/
|
||
|
|
void SendTextToOutputDevice(const char* text, SendTextCompletedHandler callback);
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Enumerate all devices and their capabilities from the installed drivers. On some operating systems,
|
||
|
|
* the ability to choose devices is limited. On windows, this will be more detailed. Call this function
|
||
|
|
* to determine what to pass to SetActiveDevice().
|
||
|
|
*/
|
||
|
|
OutputDeviceList EnumerateDevices() const;
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Sets the active driver (if there are multiple possbilities), the device to use for that driver, and the
|
||
|
|
* audio format to configure the device for. This format will also be used for calls to the Polly service.
|
||
|
|
*/
|
||
|
|
void SetActiveDevice(const std::shared_ptr<PCMOutputDriver>&, const DeviceInfo&, const CapabilityInfo&);
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Lists all available voices and their language. You can use this function to determine what to pass to the SetActiveVoice()
|
||
|
|
* function. The first member of the pair is the voice, the second is the language.
|
||
|
|
*/
|
||
|
|
Aws::Vector<std::pair<Aws::String, Aws::String>> ListAvailableVoices() const;
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Sets the active voice for use with the Polly Service.
|
||
|
|
*/
|
||
|
|
void SetActiveVoice(const Aws::String& voice);
|
||
|
|
|
||
|
|
private:
|
||
|
|
TextToSpeechManager(const std::shared_ptr<Polly::PollyClient>& pollyClient,
|
||
|
|
const std::shared_ptr<PCMOutputDriverFactory>& driverFactory);
|
||
|
|
|
||
|
|
void OnPollySynthSpeechOutcomeRecieved(const Polly::PollyClient*, const Polly::Model::SynthesizeSpeechRequest&,
|
||
|
|
const Polly::Model::SynthesizeSpeechOutcome&, const std::shared_ptr<const Aws::Client::AsyncCallerContext>&) const;
|
||
|
|
|
||
|
|
Polly::PollyClient* m_pollyClient;
|
||
|
|
std::shared_ptr<PCMOutputDriver> m_activeDriver;
|
||
|
|
Aws::Vector<std::shared_ptr<PCMOutputDriver>> m_drivers;
|
||
|
|
std::atomic<Polly::Model::VoiceId> m_activeVoice;
|
||
|
|
CapabilityInfo m_selectedCaps;
|
||
|
|
mutable std::mutex m_driverLock;
|
||
|
|
};
|
||
|
|
}
|
||
|
|
}
|