Pirros Tsiakoulis

Georgia Maniati, Alexandra Vioni, Nikolaos Ellinas et al.

In this work, we present the SOMOS dataset, the first large-scale mean opinion scores (MOS) dataset consisting of solely neural text-to-speech (TTS) samples. It can be employed to train automatic MOS prediction systems focused on the assessment of modern synthesizers, and can stimulate advancements in acoustic model evaluation. It consists of 20K synthetic utterances of the LJ Speech voice, a public domain speech dataset which is a common benchmark for building neural acoustic models and vocoders. Utterances are generated from 200 TTS systems including vanilla neural acoustic models as well as models which allow prosodic variations. An LPCNet vocoder is used for all systems, so that the samples' variation depends only on the acoustic models. The synthesized utterances provide balanced and adequate domain and length coverage. We collect MOS naturalness evaluations on 3 English Amazon Mechanical Turk locales and share practices leading to reliable crowdsourced annotations for this task. We provide baseline results of state-of-the-art MOS prediction models on the SOMOS dataset and show the limitations that such models face when assigned to evaluate TTS utterances.

Nikolaos Ellinas, Myrsini Christidou, Alexandra Vioni et al.

In this paper, we present a novel method for phoneme-level prosody control of F0 and duration using intuitive discrete labels. We propose an unsupervised prosodic clustering process which is used to discretize phoneme-level F0 and duration features from a multispeaker speech dataset. These features are fed as an input sequence of prosodic labels to a prosody encoder module which augments an autoregressive attention-based text-to-speech model. We utilize various methods in order to improve prosodic control range and coverage, such as augmentation, F0 normalization, balanced clustering for duration and speaker-independent clustering. The final model enables fine-grained phoneme-level prosody control for all speakers contained in the training set, while maintaining the speaker identity. Instead of relying on reference utterances for inference, we introduce a prior prosody encoder which learns the style of each speaker and enables speech synthesis without the requirement of reference audio. We also fine-tune the multispeaker model to unseen speakers with limited amounts of data, as a realistic application scenario and show that the prosody control capabilities are maintained, verifying that the speaker-independent prosodic clustering is effective. Experimental results show that the model has high output speech quality and that the proposed method allows efficient prosody control within each speaker's range despite the variability that a multispeaker setting introduces.

Alexandra Vioni, Georgia Maniati, Nikolaos Ellinas et al.

Current state-of-the-art methods for automatic synthetic speech evaluation are based on MOS prediction neural models. Such MOS prediction models include MOSNet and LDNet that use spectral features as input, and SSL-MOS that relies on a pretrained self-supervised learning model that directly uses the speech signal as input. In modern high-quality neural TTS systems, prosodic appropriateness with regard to the spoken content is a decisive factor for speech naturalness. For this reason, we propose to include prosodic and linguistic features as additional inputs in MOS prediction systems, and evaluate their impact on the prediction outcome. We consider phoneme level F0 and duration features as prosodic inputs, as well as Tacotron encoder outputs, POS tags and BERT embeddings as higher-level linguistic inputs. All MOS prediction systems are trained on SOMOS, a neural TTS-only dataset with crowdsourced naturalness MOS evaluations. Results show that the proposed additional features are beneficial in the MOS prediction task, by improving the predicted MOS scores' correlation with the ground truths, both at utterance-level and system-level predictions.

Konstantinos Klapsas, Nikolaos Ellinas, Karolos Nikitaras et al.

Voice cloning is a difficult task which requires robust and informative features incorporated in a high quality TTS system in order to effectively copy an unseen speaker's voice. In our work, we utilize features learned in a self-supervised framework via the Bootstrap Your Own Latent (BYOL) method, which is shown to produce high quality speech representations when specific audio augmentations are applied to the vanilla algorithm. We further extend the augmentations in the training procedure to aid the resulting features to capture the speaker identity and to make them robust to noise and acoustic conditions. The learned features are used as pre-trained utterance-level embeddings and as inputs to a Non-Attentive Tacotron based architecture, aiming to achieve multispeaker speech synthesis without utilizing additional speaker features. This method enables us to train our model in an unlabeled multispeaker dataset as well as use unseen speaker embeddings to copy a speaker's voice. Subjective and objective evaluations are used to validate the proposed model, as well as the robustness to the acoustic conditions of the target utterance.

Karolos Nikitaras, Georgios Vamvoukakis, Nikolaos Ellinas et al.

A text-to-speech (TTS) model typically factorizes speech attributes such as content, speaker and prosody into disentangled representations.Recent works aim to additionally model the acoustic conditions explicitly, in order to disentangle the primary speech factors, i.e. linguistic content, prosody and timbre from any residual factors, such as recording conditions and background noise.This paper proposes unsupervised, interpretable and fine-grained noise and prosody modeling. We incorporate adversarial training, representation bottleneck and utterance-to-frame modeling in order to learn frame-level noise representations. To the same end, we perform fine-grained prosody modeling via a Fully Hierarchical Variational AutoEncoder (FVAE) which additionally results in more expressive speech synthesis.

Konstantinos Markopoulos, Georgia Maniati, Georgios Vamvoukakis et al.

The gender of any voice user interface is a key element of its perceived identity. Recently, there has been increasing interest in interfaces where the gender is ambiguous rather than clearly identifying as female or male. This work addresses the task of generating novel gender-ambiguous TTS voices in a multi-speaker, multilingual setting. This is accomplished by efficiently sampling from a latent speaker embedding space using a proposed gender-aware method. Extensive objective and subjective evaluations clearly indicate that this method is able to efficiently generate a range of novel, diverse voices that are consistent and perceived as more gender-ambiguous than a baseline voice across all the languages examined. Interestingly, the gender perception is found to be robust across two demographic factors of the listeners: native language and gender. To our knowledge, this is the first systematic and validated approach that can reliably generate a variety of gender-ambiguous voices.

Nikolaos Ellinas, Georgios Vamvoukakis, Konstantinos Markopoulos et al.

This paper presents a method for end-to-end cross-lingual text-to-speech (TTS) which aims to preserve the target language's pronunciation regardless of the original speaker's language. The model used is based on a non-attentive Tacotron architecture, where the decoder has been replaced with a normalizing flow network conditioned on the speaker identity, allowing both TTS and voice conversion (VC) to be performed by the same model due to the inherent linguistic content and speaker identity disentanglement. When used in a cross-lingual setting, acoustic features are initially produced with a native speaker of the target language and then voice conversion is applied by the same model in order to convert these features to the target speaker's voice. We verify through objective and subjective evaluations that our method can have benefits compared to baseline cross-lingual synthesis. By including speakers averaging 7.5 minutes of speech, we also present positive results on low-resource scenarios.

Panos Kakoulidis, Nikolaos Ellinas, Georgios Vamvoukakis et al.

Existing singing voice synthesis models (SVS) are usually trained on singing data and depend on either error-prone time-alignment and duration features or explicit music score information. In this paper, we propose Karaoker, a multispeaker Tacotron-based model conditioned on voice characteristic features that is trained exclusively on spoken data without requiring time-alignments. Karaoker synthesizes singing voice and transfers style following a multi-dimensional template extracted from a source waveform of an unseen singer/speaker. The model is jointly conditioned with a single deep convolutional encoder on continuous data including pitch, intensity, harmonicity, formants, cepstral peak prominence and octaves. We extend the text-to-speech training objective with feature reconstruction, classification and speaker identification tasks that guide the model to an accurate result. In addition to multitasking, we also employ a Wasserstein GAN training scheme as well as new losses on the acoustic model's output to further refine the quality of the model.

Konstantinos Klapsas, Karolos Nikitaras, Nikolaos Ellinas et al.

A large part of the expressive speech synthesis literature focuses on learning prosodic representations of the speech signal which are then modeled by a prior distribution during inference. In this paper, we compare different prior architectures at the task of predicting phoneme level prosodic representations extracted with an unsupervised FVAE model. We use both subjective and objective metrics to show that normalizing flow based prior networks can result in more expressive speech at the cost of a slight drop in quality. Furthermore, we show that the synthesized speech has higher variability, for a given text, due to the nature of normalizing flows. We also propose a Dynamical VAE model, that can generate higher quality speech although with decreased expressiveness and variability compared to the flow based models.

Karolos Nikitaras, Konstantinos Klapsas, Nikolaos Ellinas et al.

This paper proposes an Expressive Speech Synthesis model that utilizes token-level latent prosodic variables in order to capture and control utterance-level attributes, such as character acting voice and speaking style. Current works aim to explicitly factorize such fine-grained and utterance-level speech attributes into different representations extracted by modules that operate in the corresponding level. We show that the fine-grained latent space also captures coarse-grained information, which is more evident as the dimension of latent space increases in order to capture diverse prosodic representations. Therefore, a trade-off arises between the diversity of the token-level and utterance-level representations and their disentanglement. We alleviate this issue by first capturing rich speech attributes into a token-level latent space and then, separately train a prior network that given the input text, learns utterance-level representations in order to predict the phoneme-level, posterior latents extracted during the previous step. Both qualitative and quantitative evaluations are used to demonstrate the effectiveness of the proposed approach. Audio samples are available in our demo page.

Nikolaos Ellinas, Alexandra Vioni, Panos Kakoulidis et al.

This paper introduces a cepstrum-based pitch modification method that can be applied to any mel-spectrogram representation. As a result, this method is compatible with any mel-based vocoder without requiring any additional training or changes to the model. This is achieved by directly modifying the cepstrum feature space in order to shift the harmonic structure to the desired target. The spectrogram magnitude is computed via the pseudo-inverse mel transform, then converted to the cepstrum by applying DCT. In this domain, the cepstral peak is shifted without having to estimate its position and the modified mel is recomputed by applying IDCT and mel-filterbank. These pitch-shifted mel-spectrogram features can be converted to speech with any compatible vocoder. The proposed method is validated experimentally with objective and subjective metrics on various state-of-the-art neural vocoders as well as in comparison with traditional pitch modification methods.

Michael Mitsios, Georgios Vamvoukakis, Georgia Maniati et al.

Emotion detection in textual data has received growing interest in recent years, as it is pivotal for developing empathetic human-computer interaction systems. This paper introduces a method for categorizing emotions from text, which acknowledges and differentiates between the diversified similarities and distinctions of various emotions. Initially, we establish a baseline by training a transformer-based model for standard emotion classification, achieving state-of-the-art performance. We argue that not all misclassifications are of the same importance, as there are perceptual similarities among emotional classes. We thus redefine the emotion labeling problem by shifting it from a traditional classification model to an ordinal classification one, where discrete emotions are arranged in a sequential order according to their valence levels. Finally, we propose a method that performs ordinal classification in the two-dimensional emotion space, considering both valence and arousal scales. The results show that our approach not only preserves high accuracy in emotion prediction but also significantly reduces the magnitude of errors in cases of misclassification.

Panos Kakoulidis, Nikolaos Ellinas, Georgios Vamvoukakis et al.

In this paper, we propose a singing voice synthesis model, Karaoker-SSL, that is trained only on text and speech data as a typical multi-speaker acoustic model. It is a low-resource pipeline that does not utilize any singing data end-to-end, since its vocoder is also trained on speech data. Karaoker-SSL is conditioned by self-supervised speech representations in an unsupervised manner. We preprocess these representations by selecting only a subset of their task-correlated dimensions. The conditioning module is indirectly guided to capture style information during training by multi-tasking. This is achieved with a Conformer-based module, which predicts the pitch from the acoustic model's output. Thus, Karaoker-SSL allows singing voice synthesis without reliance on hand-crafted and domain-specific features. There are also no requirements for text alignments or lyrics timestamps. To refine the voice quality, we employ a U-Net discriminator that is conditioned on the target speaker and follows a Diffusion GAN training scheme.

Alexandra Vioni, Myrsini Christidou, Nikolaos Ellinas et al.

This paper presents a method for controlling the prosody at the phoneme level in an autoregressive attention-based text-to-speech system. Instead of learning latent prosodic features with a variational framework as is commonly done, we directly extract phoneme-level F0 and duration features from the speech data in the training set. Each prosodic feature is discretized using unsupervised clustering in order to produce a sequence of prosodic labels for each utterance. This sequence is used in parallel to the phoneme sequence in order to condition the decoder with the utilization of a prosodic encoder and a corresponding attention module. Experimental results show that the proposed method retains the high quality of generated speech, while allowing phoneme-level control of F0 and duration. By replacing the F0 cluster centroids with musical notes, the model can also provide control over the note and octave within the range of the speaker.

Myrsini Christidou, Alexandra Vioni, Nikolaos Ellinas et al.

This paper presents a method for phoneme-level prosody control of F0 and duration on a multispeaker text-to-speech setup, which is based on prosodic clustering. An autoregressive attention-based model is used, incorporating multispeaker architecture modules in parallel to a prosody encoder. Several improvements over the basic single-speaker method are proposed that increase the prosodic control range and coverage. More specifically we employ data augmentation, F0 normalization, balanced clustering for duration, and speaker-independent prosodic clustering. These modifications enable fine-grained phoneme-level prosody control for all speakers contained in the training set, while maintaining the speaker identity. The model is also fine-tuned to unseen speakers with limited amounts of data and it is shown to maintain its prosody control capabilities, verifying that the speaker-independent prosodic clustering is effective. Experimental results verify that the model maintains high output speech quality and that the proposed method allows efficient prosody control within each speaker's range despite the variability that a multispeaker setting introduces.

Konstantinos Markopoulos, Nikolaos Ellinas, Alexandra Vioni et al.

In this paper, a text-to-rapping/singing system is introduced, which can be adapted to any speaker's voice. It utilizes a Tacotron-based multispeaker acoustic model trained on read-only speech data and which provides prosody control at the phoneme level. Dataset augmentation and additional prosody manipulation based on traditional DSP algorithms are also investigated. The neural TTS model is fine-tuned to an unseen speaker's limited recordings, allowing rapping/singing synthesis with the target's speaker voice. The detailed pipeline of the system is described, which includes the extraction of the target pitch and duration values from an a capella song and their conversion into target speaker's valid range of notes before synthesis. An additional stage of prosodic manipulation of the output via WSOLA is also investigated for better matching the target duration values. The synthesized utterances can be mixed with an instrumental accompaniment track to produce a complete song. The proposed system is evaluated via subjective listening tests as well as in comparison to an available alternate system which also aims to produce synthetic singing voice from read-only training data. Results show that the proposed approach can produce high quality rapping/singing voice with increased naturalness.

Georgia Maniati, Nikolaos Ellinas, Konstantinos Markopoulos et al.

The idea of using phonological features instead of phonemes as input to sequence-to-sequence TTS has been recently proposed for zero-shot multilingual speech synthesis. This approach is useful for code-switching, as it facilitates the seamless uttering of foreign text embedded in a stream of native text. In our work, we train a language-agnostic multispeaker model conditioned on a set of phonologically derived features common across different languages, with the goal of achieving cross-lingual speaker adaptation. We first experiment with the effect of language phonological similarity on cross-lingual TTS of several source-target language combinations. Subsequently, we fine-tune the model with very limited data of a new speaker's voice in either a seen or an unseen language, and achieve synthetic speech of equal quality, while preserving the target speaker's identity. With as few as 32 and 8 utterances of target speaker data, we obtain high speaker similarity scores and naturalness comparable to the corresponding literature. In the extreme case of only 2 available adaptation utterances, we find that our model behaves as a few-shot learner, as the performance is similar in both the seen and unseen adaptation language scenarios.

Nikolaos Ellinas, Georgios Vamvoukakis, Konstantinos Markopoulos et al.

This paper presents an end-to-end text-to-speech system with low latency on a CPU, suitable for real-time applications. The system is composed of an autoregressive attention-based sequence-to-sequence acoustic model and the LPCNet vocoder for waveform generation. An acoustic model architecture that adopts modules from both the Tacotron 1 and 2 models is proposed, while stability is ensured by using a recently proposed purely location-based attention mechanism, suitable for arbitrary sentence length generation. During inference, the decoder is unrolled and acoustic feature generation is performed in a streaming manner, allowing for a nearly constant latency which is independent from the sentence length. Experimental results show that the acoustic model can produce feature sequences with minimal latency about 31 times faster than real-time on a computer CPU and 6.5 times on a mobile CPU, enabling it to meet the conditions required for real-time applications on both devices. The full end-to-end system can generate almost natural quality speech, which is verified by listening tests.