Add multilingual support to ForwardTTS

[NeonBohdan]

Multilingual for now is only initialized in VITS

TTS/TTS/tts/models/vits.py

Lines 589 to 592 in c44e39d

	super().__init__(config, ap, tokenizer, speaker_manager, language_manager)
	self.init_multispeaker(config)
	self.init_multilingual(config)

But SpeedySpeech has some pros over glowTTS(part of VITS)

• faster (x2-x3)
• smaller (100MiB->25MiB)

But when I want to train multilingual model I need to get back to VITS
Can this feature be extended to SpeedySpeech?

TTS/TTS/tts/models/forward_tts.py

Lines 181 to 185 in c44e39d

	super().__init__(config, ap, tokenizer, speaker_manager)
	self._set_model_args(config)
	self.init_multispeaker(config)

[erogol]

We are currently trying a model for multi-speaker datasets. Next, we can step into multilingual but it has not been experimented with yet.