It has long been speculated whether communication between humans and machines based on natural speech related cortical activity is possible. Over the past decade, studies have suggested that it is feasible to recognize isolated aspects of speech from neural signals, such as auditory features, phones or one of a few isolated words. However, until now it remained an unsolved challenge to decode continuously spoken speech from the neural substrate associated with speech and language processing. Here, we show for the first time that continuously spoken speech can be decoded into the expressed words from intracranial electrocorticographic (ECoG) recordings. Specifically, we implemented a system, which we call Brain-To-Text that models single phones, employs techniques from automatic speech recognition (ASR), and thereby transforms brain activity while speaking into the corresponding textual representation. Our results demonstrate that our system achieved word error rates as low as 25% and phone error rates below 50%. Additionally, our approach contributes to the current understanding of the neural basis of continuous speech production by identifying those cortical regions that hold substantial information about individual phones. In conclusion, the Brain-To-Text system described in this paper represents an important step towards human-machine communication based on imagined speech.

When interacting with technical systems, users experience mental workload. Particularly in multitasking scenarios (e.g. interacting with the car navigation system while driving) it is desired to not distract the users from their primary task. For such purposes, human-machine interfaces (HCIs) are desirable which continuously monitor the users' workload and dynamically adapt the behavior of the interface to the measured workload. While memory tasks have been shown to illicit hemodynamic responses in the brain when averaging over multiple trials, a robust single trial classification is a crucial prerequisite for the purpose of dynamically adapting HCIs to the workload of its user.The prefrontal cortex (PFC) plays an important role in the processing of memory and the associated workload. In this study of 10 subjects, we used functional Near-Infrared Spectroscopy (fNIRS), a non-invasive imaging modality, to sample workload activity in the PFC. The results show up to 78% accuracy for single-trial discrimination of three levels of workload from each other. We use an n-back task (n ∈ {1, 2, 3}) to induce different levels of workload, forcing subjects to continuously remember the last one, two or three of rapidly changing items.Our experimental results show that measuring hemodynamic responses in the PFC with fNIRS, can be used to robustly quantify and classify mental workload.Single trial analysis is still a young field that suffers from a general lack of standards. To increase comparability of fNIRS methods and results, the data corpus for this study is made available online.

Objective . Direct synthesis of speech from neural signals could provide a fast and natural way of communication to people with neurological diseases. Invasively-measured brain activity (electrocorticography; ECoG) supplies the necessary temporal and spatial resolution to decode fast and complex processes such as speech production. A number of impressive advances in speech decoding using neural signals have been achieved in recent years, but the complex dynamics are still not fully understood. However, it is unlikely that simple linear models can capture the relation between neural activity and continuous spoken speech. Approach . Here we show that deep neural networks can be used to map ECoG from speech production areas onto an intermediate representation of speech (logMel spectrogram). The proposed method uses a densely connected convolutional neural network topology which is well-suited to work with the small amount of data available from each participant. Main results . In a study with six participants, we achieved correlations up to r = 0.69 between the reconstructed and original logMel spectrograms. We transfered our prediction back into an audible waveform by applying a Wavenet vocoder. The vocoder was conditioned on logMel features that harnessed a much larger, pre-existing data corpus to provide the most natural acoustic output. Significance . To the best of our knowledge, this is the first time that high-quality speech has been reconstructed from neural recordings during speech production using deep neural networks.