Speaker 1: Hi guys, welcome to another episode of Learn Technology in 5 Minutes by MAKERDEMY. This is Saharsh and I am part of the instructor team at MAKERDEMY. In this episode, we will learn about the Azure Speech Services. Azure Speech Services are used to enable speech processing in apps. Azure Speech Services enables us to convert speech to text and text to speech while also translating text to other languages and identifying speakers. These services are provided as APIs to easily integrate them into our applications by giving an API call to Azure. Services such as these are also used in hands-free tools like Amazon Alexa or Google Assistant. Azure Speech Services provide the following APIs. Speech-to-Text API Text-to-Speech API Speech Translation API Speaker Recognition API Speaker Verification API Let's learn about each of the APIs individually and understand their capabilities and benefits. Speech-to-Text API Speech-to-Text, commonly referred to as Speech Recognition, enables real-time translation of audio streams into text. In simple words, audio is the input and text is the output. The most common use case of speech-to-text that we can think of is subtitles. YouTube uses this technology to provide auto-generated subtitles to a video. This service provides massive help to content creators and authors. It is powered by the same recognition technology that Microsoft uses for Cortana. It can also integrate with the translation and text-to-speech service offerings. It also supports multiple languages. Text-to-Speech API Text-to-Speech, in simple words, takes text as input and gives audio as output. This technology is commonly observed in mobile phone chatbots like Siri in iOS and Google Assistant in Android. Writers and authors can use this service to convert their ebooks into audiobooks. There are many features that this API offers. If you want to check them out, please visit the link given in the video description. Speech Translation API Speech Translation service enables real-time speech-to-speech and speech-to-text translation of audio streams. With the Speech Software Development Kit, we can give our applications and devices access to speech translation between multiple languages and multiple accents of the same language. This service is widely used in customer care services. Core features associated with speech translation are Speech-to-text translation with recognition results Speech-to-speech translation Support for translation to multiple target languages Interim recognition and translation results Speaker Recognition API The Speaker Recognition service provides algorithms that can identify speakers by their unique voice characteristics using voice biometry. Speaker recognition answers the question, who is speaking? We need to provide audio training data for a single speaker that creates a profile based on the unique characteristics of the speaker's voice. We can then cross-check with voice samples against this profile to verify that the speaker is the same person or cross-check the voice samples against a group of enrolled speaker profiles to check if it matches any profile in the group. This service is used in various commercial applications like biometric authentication, phone banking, voicemail, and forensics. This service is still in its preview state and its full release on Azure is awaited. Speaker Verification API This service can be used to authenticate individuals for secure customer interactions in a wide range of solutions like customer identity verification in call centers and contactless facility access. Let's see how speaker recognition works. Speaker verification is of two types – text-dependent or text-independent. For text-dependent verification, the speaker's voice is registered by saying a phrase from a set of predefined passphrases. Voice features are extracted from this audio to form a unique voice signature while the chosen phrase is also recognized. Both the passphrase and the voice signature are used for verification here. Text-independent verification has no restrictions on what to say while registering the speaker's voice as it only extracts voice features to compute similarity. These services are used by highly secured government agencies where multi-factor authentication is a must. But Azure clearly mentions that this API is not intended to determine if the audio is taken from a live person or if it is taken from a recording of an enrolled speaker. All these APIs that we have discussed so far are a part of the speech services that the Azure Cognitive Services provide. That is all, folks. If you like this video, smash the subscribe button and ring the bell to be updated about our future video releases.