You want to use supervised learning to build a speech recognition system. The figure above suggests that in order for a neural network (deep learning) to achieve the best performance, you would ideally use: (Select all that apply)1 pointA large dataset (of audio files and the corresponding text transcript)A small dataset (of audio files and the corresponding text transcript)A large neural networkA small neural network

What’s the most efficient way to transcribe speech?You can use a speech API.All options are correct.Use a Dictionary website for a partial transcription, then using ML to fill in what’s missing.You can collect audio data, train it and predict with it.

Which of these activities is NOT required in order for a neural network to synthesize human voice?1 pointDeconstruct sentences to decipher the context of useGenerate audio data and run it through the network to see if it validates it as belonging to the subjectContinue to correct the sample and run it through the classifier, repetitively, till an accurate voice sample is created Ingest numerous samples of a person’s voice until it can tell whether a new voice sample belongs to the same person

Question 9Which of these activities is NOT required in order for a neural network to synthesize human voice?1 pointContinue to correct the sample and run it through the classifier, repetitively, till an accurate voice sample is created Ingest numerous samples of a person’s voice until it can tell whether a new voice sample belongs to the same personGenerate audio data and run it through the network to see if it validates it as belonging to the subjectDeconstruct sentences to decipher the context of use

6. Some linguistics researchers want to train an AI system to recognize mispronunciation of certain words in Mandarin: as the user pronounces one of those word in Mandarin, the system will be able to indicate in which one of the four Mandarin intonations the word is pronounced. This will then allow the user to check if the correct intonation is pronounced. This means the system will need to process words that are pronounced in the wrong intonation. What kind of audio recordings should be the most suitable for the researchers to collect for training this system?一些語言學研究員希望訓練一個能辨認某些普通話誤讀字的人工智能系統：當用家以普通話讀出這些單字的其中一個，這個系統便能指出這個讀音是用了四個普通話聲調中的哪一個，用家就能以此檢查是否讀出了正確的聲調。這也代表這個系統需要處理用錯誤的聲調讀出的單字。哪些錄音數據最適合研究員們收集來訓練這個系統?Words pronounced under the same intonation in Mandarin 以同一普通話聲調發音的單字Correct Mandarin pronunciations of words 單字的正確普通話讀音Random paragraphs read by different people in Mandarin 不同人士用普通話朗讀的、隨意挑選的段落Words pronounced in Mandarin in each intonation regardless of whether the correct intonations are used用每一個普通話聲調讀出的單字，不管是否用了正確的聲調

To which of these tasks would you apply a many-to-one RNN architecture?Question 7Answera.   Both sentiment classification and gender recognition from speechb.Gender recognition from speech (input an audio clip and output a label indicating the speaker’s gender)c.   Speech recognition (input an audio clip and output a transcript)d. Sentiment classification (input a piece of text and output a 0/1 to denote positive or negative sentiment)