Abstract-Much is known about the design of automated systems to search broadcast news, but it has only recently become possible to apply similar techniques to large collections of spontaneous speech. This paper presents initial results from experiments with speech recognition, topic segmentation, topic categorization, and named entity detection using a large collection of recorded oral histories. The work leverages a massive manual annotation effort on 10 000 h of spontaneous speech to evaluate the degree to which automatic speech recognition (ASR)-based segmentation and categorization techniques can be adapted to approximate decisions made by human annotators. ASR word error rates near 40% were achieved for both English and Czech for heavily accented, emotional and elderly spontaneous speech based on 65-84 h of transcribed speech. Topical segmentation based on shifts in the recognized English vocabulary resulted in 80% agreement with manually annotated boundary positions at a 0.35 false alarm rate. Categorization was considerably more challenging, with a nearestneighbor technique yielding F = 0 3. This is less than half the value obtained by the same technique on a standard newswire categorization benchmark, but replication on human-transcribed interviews showed that ASR errors explain little of that difference. The paper concludes with a description of how these capabilities Manuscript

This paper presents part of the data collection efforts undergone within the project COMPANIONS whose aim is to develop a set of dialogue systems that will be able to act as an artificial "companions" for human users. One of these systems, being developed in Czech language, is designed to be a partner of elderly people which will be able to talk with them about the photographs that capture mostly their family memories. The paper describes in detail the collection of natural dialogues using the Wizard of Oz scenario and also the re-use of the collected data for the creation of the expressive speech corpus that is planned for the development of the limited-domain Czech expressive TTS system.

The main objective of the work presented in this paper was to develop a complete system that would accomplish the original visions of the MALACH project. Those goals were to employ automatic speech recognition and information retrieval techniques to provide improved access to the large video archive containing recorded testimonies of the Holocaust survivors. The system has been so far developed for the Czech part of the archive only. It takes advantage of the state-of-the-art speech recognition system tailored to the challenging properties of the recordings in the archive (elderly speakers, spontaneous speech and emotionally loaded content) and its close coupling with the actual search engine. The design of the algorithm adopting the spoken term detection approach is focused on the speed of the retrieval. The resulting system is able to search through the 1,000 h of video constituting the Czech portion of the archive and find query word occurrences in the matter of seconds. The phonetic search implemented alongside the search based on the lexicon words allows to find even the words outside the ASR system lexicon such as names, geographic locations or Jewish slang.

In this paper, we present a novel method for term score estimation. The method is primarily designed for scoring the out-of-vocabulary terms, however it could also estimate scores for in-vocabulary results. The term score is computed as a cosine distance of two pronunciation embeddings. The first one is generated from the grapheme representation of the searched term, while the second one is computed from the recognized phoneme confusion network. The embeddings are generated by specifically trained recurrent neural network built on the idea of Siamese neural networks. The RNN is trained from recognition results on word-and phone-level in an unsupervised fashion without need of any hand-labeled data. The method is evaluated on the MALACH data in two languages, English and Czech. The results are compared with two baseline methods for OOV term detection.

No abstract