Text-to-speech models trained on large-scale datasets have demonstrated impressive in-context learning capabilities and naturalness. However, control of speaker identity and style in these models typically requires conditioning on reference speech recordings, limiting creative applications. Alternatively, natural language prompting of speaker identity and style has demonstrated promising results and provides an intuitive method of control. However, reliance on human-labeled descriptions prevents scaling to large datasets. Our work bridges the gap between these two approaches. We propose a scalable method for labeling various aspects of speaker identity, style, and recording conditions. We then apply this method to a 45k hour dataset, which we use to train a speech language model. Furthermore, we propose simple methods for increasing audio fidelity, significantly outperforming recent work despite relying entirely on found data. Our results demonstrate high-fidelity speech generation in a diverse range of accents, prosodic styles, channel conditions, and acoustic conditions, all accomplished with a single model and intuitive natural language conditioning. Audio samples can be heard at https://text-description-to-speech.com/.
Machine learning approaches to modelling analog audio effects have seen intensive investigation in recent years, particularly in the context of non-linear time-invariant effects such as guitar amplifiers. For modulation effects such as phasers, however, new challenges emerge due to the presence of the low-frequency oscillator which controls the slowly time-varying nature of the effect. Existing approaches have either required foreknowledge of this control signal, or have been non-causal in implementation. This work presents a differentiable digital signal processing approach to modelling phaser effects in which the underlying control signal and time-varying spectral response of the effect are jointly learned. The proposed model processes audio in short frames to implement a time-varying filter in the frequency domain, with a transfer function based on typical analog phaser circuit topology. We show that the model can be trained to emulate an analog reference device, while retaining interpretable and adjustable parameters. The frame duration is an important hyper-parameter of the proposed model, so an investigation was carried out into its effect on model accuracy. The optimal frame length depends on both the rate and transient decay-time of the target effect, but the frame length can be altered at inference time without a significant change in accuracy.
Appropriate prosody is critical for successful spoken communication. Contextual word embeddings are proven to be helpful in predicting prosody but do not allow for choosing between plausible prosodic renditions. Reference-based TTS models attempt to address this by conditioning speech generation on a reference speech sample. These models can generate expressive speech but this requires finding an appropriate reference. Sufficiently large generative language models have been used to solve various language-related tasks. We explore whether such models can be used to suggest appropriate prosody for expressive TTS. We train a TTS model on a non-expressive corpus and then prompt the language model to suggest changes to pitch, energy and duration. The prompt can be designed for any task and we prompt the model to make suggestions based on target speaking style and dialogue context. The proposed method is rated most appropriate in 49.9\% of cases compared to 31.0\% for a baseline model.
Generating expressive speech with rich and varied prosody continues to be a challenge for Text-to-Speech. Most efforts have focused on sophisticated neural architectures intended to better model the data distribution. Yet, in evaluations it is generally found that no single model is preferred for all input texts. This suggests an approach that has rarely been used before for Text-to-Speech: an ensemble of models. We apply ensemble learning to prosody prediction. We construct simple ensembles of prosody predictors by varying either model architecture or model parameter values. To automatically select amongst the models in the ensemble when performing Text-to-Speech, we propose a novel, and computationally trivial, variance-based criterion. We demonstrate that even a small ensemble of prosody predictors yields useful diversity, which, combined with the proposed selection criterion, outperforms any individual model from the ensemble.
Some recent models for Text-to-Speech synthesis aim to transfer the prosody of a reference utterance to the generated target synthetic speech. This is done by using a learned embedding of the reference utterance, which is used to condition speech generation. During training, the reference utterance is identical to the target utterance. Yet, during synthesis, these models are often used to transfer prosody from a reference that differs from the text or speaker being synthesized. To address this inconsistency, we propose to use a different, but prosodically-related, utterance during training too. We believe this should encourage the model to learn to transfer only those characteristics that the reference and target have in common. If prosody transfer methods do indeed transfer prosody they should be able to be trained in the way we propose. However, results show that a model trained under these conditions performs significantly worse than one trained using the target utterance as a reference. To explain this, we hypothesize that prosody transfer models do not learn a transferable representation of prosody, but rather an utterance-level representation which is highly dependent on both the reference speaker and reference text.
Most state-of-the-art Text-to-Speech systems use the mel-spectrogram as an intermediate representation, to decompose the task into acoustic modelling and waveform generation. A mel-spectrogram is extracted from the waveform by a simple, fast DSP operation, but generating a high-quality waveform from a mel-spectrogram requires computationally expensive machine learning: a neural vocoder. Our proposed ``autovocoder'' reverses this arrangement. We use machine learning to obtain a representation that replaces the mel-spectrogram, and that can be inverted back to a waveform using simple, fast operations including a differentiable implementation of the inverse STFT. The autovocoder generates a waveform 5 times faster than the DSP-based Griffin-Lim algorithm, and 14 times faster than the neural vocoder HiFi-GAN. We provide perceptual listening test results to confirm that the speech is of comparable quality to HiFi-GAN in the copy synthesis task.
Text does not fully specify the spoken form, so text-to-speech models must be able to learn from speech data that vary in ways not explained by the corresponding text. One way to reduce the amount of unexplained variation in training data is to provide acoustic information as an additional learning signal. When generating speech, modifying this acoustic information enables multiple distinct renditions of a text to be produced. Since much of the unexplained variation is in the prosody, we propose a model that generates speech explicitly conditioned on the three primary acoustic correlates of prosody: $F_{0}$, energy and duration. The model is flexible about how the values of these features are specified: they can be externally provided, or predicted from text, or predicted then subsequently modified. Compared to a model that employs a variational auto-encoder to learn unsupervised latent features, our model provides more interpretable, temporally-precise, and disentangled control. When automatically predicting the acoustic features from text, it generates speech that is more natural than that from a Tacotron 2 model with reference encoder. Subsequent human-in-the-loop modification of the predicted acoustic features can significantly further increase naturalness.
Text-to-speech is now able to achieve near-human naturalness and research focus has shifted to increasing expressivity. One popular method is to transfer the prosody from a reference speech sample. There have been considerable advances in using prosody transfer to generate more expressive speech, but the field lacks a clear definition of what successful prosody transfer means and a method for measuring it. We introduce a dataset of prosodically-varied reference natural speech samples for evaluating prosody transfer. The samples include global variations reflecting emotion and interpersonal attitude, and local variations reflecting topical emphasis, propositional attitude, syntactic phrasing and marked tonicity. The corpus only includes prosodic variations that listeners are able to distinguish with reasonable accuracy, and we report these figures as a benchmark against which text-to-speech prosody transfer can be compared. We conclude the paper with a demonstration of our proposed evaluation methodology, using the corpus to evaluate two text-to-speech models that perform prosody transfer.
Many speech synthesis datasets, especially those derived from audiobooks, naturally comprise sequences of utterances. Nevertheless, such data are commonly treated as individual, unordered utterances both when training a model and at inference time. This discards important prosodic phenomena above the utterance level. In this paper, we leverage the sequential nature of the data using an acoustic context encoder that produces an embedding of the previous utterance audio. This is input to the decoder in a Tacotron 2 model. The embedding is also used for a secondary task, providing additional supervision. We compare two secondary tasks: predicting the ordering of utterance pairs, and predicting the embedding of the current utterance audio. Results show that the relation between consecutive utterances is informative: our proposed model significantly improves naturalness over a Tacotron 2 baseline.
In English, prosody adds a broad range of information to segment sequences, from information structure (e.g. contrast) to stylistic variation (e.g. expression of emotion). However, when learning to control prosody in text-to-speech voices, it is not clear what exactly the control is modifying. Existing research on discrete representation learning for prosody has demonstrated high naturalness, but no analysis has been performed on what these representations capture, or if they can generate meaningfully-distinct variants of an utterance. We present a phrase-level variational autoencoder with a multi-modal prior, using the mode centres as "intonation codes". Our evaluation establishes which intonation codes are perceptually distinct, finding that the intonation codes from our multi-modal latent model were significantly more distinct than a baseline using k-means clustering. We carry out a follow-up qualitative study to determine what information the codes are carrying. Most commonly, listeners commented on the intonation codes having a statement or question style. However, many other affect-related styles were also reported, including: emotional, uncertain, surprised, sarcastic, passive aggressive, and upset.