Викторина, связанная с руководством · Средний Уровень
Тест на искусственный интеллект в колл-центре по речевому анализу
Протестируйте распознавание речи, диаризацию и анализ настроений в исследованиях колл-центров, включая ограничения и объем каждого цитируемого исследования.
Вопрос 1 из 8
What output can automatic speech recognition provide for speech analytics?
Предварительный просмотр каждого вопроса8 вопросы
Ответы и пояснения появляются по мере прохождения теста.
- What output can automatic speech recognition provide for speech analytics?
- A machine-generated transcript for downstream text analysis
- A guaranteed speaker identity
- A final diagnosis of the caller’s emotional state
- A payment authorization
- What does speaker diarization do in a call recording?
- Переводит звонок на другой язык
- Segments the recording into speaker turns and assigns speaker labels
- Измеряет, насколько громким был звонок
- Шифрует запись
- In IBM’s two-stream call-center emotion-recognition study, which information streams are combined?
- Video and keystroke timing
- Customer purchase history and account balance
- Acoustic and lexical information
- Only transcript punctuation
- Какая акустическая функция помогает определить эмоциональное состояние клиента, помимо используемых слов?
- Размер файла записи
- Идентификатор сотрудника агента
- Телефонный оператор
- Высота, темп и громкость (просодия)
- In a two-stream call-center emotion system, what do acoustic features contribute?
- Evidence from vocal delivery, such as how speech sounds
- The caller’s verified identity
- The account’s payment history
- A guaranteed diagnosis of the caller’s emotion
- What task did Vidrascu and Devillers study for customer-service phone calls?
- Predicting credit scores from call recordings
- Speaker segmentation and emotion-related analysis of conversational turns
- Generating live agent sales recommendations
- Automating legal decisions about customer accounts
- Why might a call-center sentiment system combine lexical and acoustic features?
- Acoustic features replace the transcript in every task
- Lexical features reveal the phone carrier
- The words and the way they are spoken provide complementary signals
- Combining them guarantees the caller’s true emotion is known
- What does word-error rate compare in speech recognition evaluation?
- Speaker identity against a customer account
- Emotion labels against call duration
- Call volume against agent schedule
- Substitutions, deletions and insertions against the words in a reference transcript