Text-to-speech
2 TopicsAI Assistant: Producing Highly Realistic Audio
As a course author, you want to do more than just present information—you want to create multi-sensory e-learning experiences that resonate with learners. Using sound creatively can help you get there. AI Assistant’s text-to-speech and sound effects features let you create highly realistic AI-generated voices and sound effects for more immersive and accessible content. Available only in Storyline 360, these features can be accessed from the Insert Audio dropdown in the AI Assistant menu within the ribbon. Find them under the Home or Insert tab when you’re in slide view or from the AI Assistant side panel as quick action buttons for added convenience. Bring Narration to Life with AI-generated Voices If you’ve ever used classic text-to-speech, you probably wished the voices sounded less, well, robotic. AI Assistant’s text-to-speech brings narration to life with contextually aware AI-generated voices that sound more natural—and human! Check out the difference in quality between a standard voice, neural voice, and AI-generated voice by playing the text-to-speech examples below. Standard Voice Your browser does not support the audio element. Neural Voice Your browser does not support the audio element. AI-generated Voice Your browser does not support the audio element. To get started, click the Insert Audio icon in the AI Assistant menu to open the Generate AI Audio dialog box. A library of AI-generated voices—which you can filter by Gender, Age, and Accent—displays under the Voices tab. The voices also have descriptions like “deep,” “confident,” “crisp,” “intense,” and “soothing” and categories that can help you determine their ideal use cases, from news broadcasts to meditation, or even ASMR. Find these qualities under the voice’s name, and use the play button to preview the voice. Currently, there are 52 pre-made voices to choose from, and you can mark your favorites by clicking the heart icon. This way, you can easily access your preferred voices without having to scroll through the list. Toggle the View option to Favorites to find all your favorite voices, or In project to see voices used in the current project. Once you’ve decided on a voice, click the button labeled Use to switch to the Text-to-Speech tab. Your chosen voice is already pre-selected. Next, enter your script in the text box provided or click the add from slide notes link to copy notes from your slide. The script can be a maximum of 5,000 characters. For accessibility, select Generate closed captions—AI Assistant will generate closed captions automatically. Adjust the Voice Settings Unlike classic text-to-speech, the AI-generated voices in AI Assistant’s text-to-speech can be customized for a tailored voice performance. The Model setting lets you choose between the Multilingual v2 model—highly stable, exceptionally accurate, lifelike speech with support for 29 languages—and Turbo v2.5 model—slightly less stable but 300% faster with support for 32 languages. Play the following samples to listen and compare the voices generated by each model. Multilingual v2 Your browser does not support the audio element. Turbo v2.5 Your browser does not support the audio element. The setting forStability controls the balance between the voice’s steadiness and randomness. The Similarity setting determines how closely the AI should adhere to the original voice when attempting to replicate it. The defaults are set to 0.50 for the stability slider and 0.75 for the similarity slider, but you can play around with these settings to find the right balance for your content. Additional settings include Style exaggeration, which amplifies the style of the original voice, and Speaker boost, which enhances the similarity between synthesized speech and the voice. Note that if either of those settings is adjusted, generating your speech will take longer. Do I Need to Use SSML? AI Assistant has limited support for speech synthesis markup language (SSML) because AI-generated voices are designed to understand the relationship between words and adjust delivery accordingly. If you need to manually control pacing, you can add a pause. The most consistent way to do that is by inserting the syntax <break time="1.5s" /> into your script. This creates an exact and natural pause in the speech. For example: With their keen senses <break time="1.5s" /> cats are skilled hunters. Use seconds to describe a break of up to three seconds in length. You can try a simple dash - or em-dash — to insert a brief pause or multiple dashes for a longer pause. Ellipsis ... will also sometimes work to add a pause between words. However, these options may not work consistently, so we recommend using the syntax above for consistency. Just keep in mind that an excessive number of break tags can potentially cause instability. Multilingual Voices Expand Your Reach Another compelling benefit of AI-generated text-to-speech is the ability to bridge language gaps, allowing you to connect with international audiences. With support for up to 32 languages depending on the model—including some with multiple accents and dialects—AI Assistant’s text-to-speech helps your content resonate with a global audience. All you have to do is type or paste your script in the supported language you want AI Assistant to use. (Even though the voice description notes a specific accent or language, AI Assistant will generate the narration in the language used in your script.) Note that some voices tend to work best with certain accents or languages, so feel free to experiment with different voices to find the best fit for your needs. The table below provides a quick rundown of supported languages. Available in Multilingual v2 and Turbo v2.5: English (USA) English (UK) English (Australia) English (Canada) Japanese Chinese German Hindi French (France) French (Canada) Korean Portuguese (Brazil) Portuguese (Portugal) Italian Spanish (Spain) Spanish (Mexico) Indonesian Dutch Turkish Filipino Polish Swedish Bulgarian Romanian Arabic (Saudi Arabia) Arabic (UAE) Czech Greek Finnish Croatian Malay Slovak Danish Tamil Ukrainian Russian Available only in Turbo v2.5: Hungarian Norwegian Vietnamese Create Sound Effects Using Prompts Sound effects that align with your theme and content can highlight important actions or feedback, like clicking a button or choosing a correct answer, offering a more engaging and effective e-learning experience. With AI Assistant’s sound effects, you can now use prompts to easily create nearly any sound imaginable. No more wasting time scouring the web for pre-made sounds that may cost extra! Start creating high-quality sound effects by going to the AI Assistant menu in the ribbon under the Home or Insert tab. Then, click the lower half of the Insert Audio icon, and choose Sound Effects. (You can also access it from the Audio dropdown within the Insert tab. Simply select Sound Effects under the AI Audio option.) In the text box, describe the sound effect you want and choose a duration. You can adjust the Prompt influence slider to give AI Assistant more or less creative license in generating the sound. Since AI Assistant understands natural language, sound effects can be created using anything from a simple prompt like “a single mouse click” to a very complex one that describes multiple sounds or a sequence of sounds in a specific order. Just note you have a maximum of 450 characters to describe the sound you want to generate. Play the following audio samples to listen to sound effects created using a simple prompt and a complex one. Your browser does not support the audio element. Prompt: A single mouse click Your browser does not support the audio element. Prompt: Dogs barking, then lightning strikes You can also adjust the Duration—how long the sound effect plays—up to a maximum of 22 seconds. For example, if your prompt is “barking dog” and you set the duration to 10 seconds, you’ll get continuous barking, but a duration of two seconds is one quick bark. Adjusting the Prompt Influence slider to the right makes AI Assistant strictly adhere to your prompt, while sliding it to the left allows more free interpretation. Some Pro Terms to Know Using audio terminology—specialized vocabulary that audio experts use in their work—can help improve your prompts and produce even more dynamic sound effects. Here are a few examples: Braam: A deep, resonant, and often distorted bass sound used in media, particularly in trailers, to create a sense of tension, power, or impending doom. Whoosh: A quick, swooshing sound often used to emphasize fast motion, transitions, or dramatic moments. Impact: A sharp, striking noise used to signify a collision, hit, or sudden forceful contact, often to highlight a moment of action or emphasis. Glitch: A short, jarring, and usually digital noise that mimics a malfunction or distortion, commonly used to convey errors. Foley: The process of recreating and recording everyday sound effects like movements and object sounds in sync with the visuals of a film, videos, or other media. Here’s something fun to try! Generate a 3-second sound effect using the prompt “studio quality, sound designed whoosh and braam impact.” Increasing the duration may produce better sound effects but will also create more dead air towards the end. Pro tip: Onomatopoeias—words like “buzz,” “boom,” “click,” and “pop” that imitate natural sounds—are also important sound effects terms. Use them in your prompts to create more realistic sound effects. Video Tutorials Want to learn more before getting started? Check out our video tutorials for additional guidance on using AI Assistant to generate text-to-speech and sound effects. Create AI-generated Text-to-Speech Create AI-generated Sound Effects Articulate 360 Training also has additional video tutorials on using other AI Assistant features. Use AI Assistant features in Rise 360 Use AI Assistant features in Storyline 360 You must be logged in to your Articulate 360 account to watch the videos. Don’t have an account yet? Sign up for a free trial now!3.6KViews10likes0CommentsStoryline 360: Enrich Audio Narrations with Classic or AI-Generated Text-to-Speech
Course authors have long relied on the classic text-to-speech feature in Storyline 360 to create quick audio narration for their e-learning content and speed up course development. However, even with the neural voice options that have been added to the standard ones, the voices in the classic text-to-speech feature can sound robotic, making for a less natural and engaging learner experience. Now, AI-generated text-to-speech is changing the game. The newest addition to your authoring toolkit, AI Assistant’s text-to-speech gives you access to incredibly lifelike, AI-generated voices that are hard to distinguish from a real human voice. So will you keep using the classic version or embrace the brave new world of AI text-to-speech technology? Let’s take a look at the pros and cons of each to help you decide which option to choose for your next project. Classic Text-to-Speech Storyline’s classic text-to-speech has evolved significantly over the years. In particular, the introduction of neural voices empowered authors to create more realistic and natural-sounding narrations. Here’s an overview of when classic text-to-speech may be the best choice to elevate your audio content. You need certain languages. Classic text-to-speech allows you to create narrations for diverse audiences with support for multiple languages in standard and neural voice. In addition, some of these languages—including Icelandic, Welsh, Catalan, and Irish—are only currently available in classic text-to-speech. You need full SSML support. Unlike AI-generated text-to-speech, classic text-to-speech offers full speech synthesis markup language (SSML) support. This allows you to fine-tune narrations by adjusting the speaking rate, modifying pronunciation, adding pauses, and more to boost clarity and interest. You have reservations about using AI. Not everyone is ready to embrace new technologies like generative AI, and even some organizations have restrictions on using AI-powered tools. Outside of these specific circumstances, however, the overall quality of classic text-to-speech voices may not be sufficiently natural, especially for more complex or nuanced content. Want to judge for yourself? You can hear classic text-to-speech in action by playing the narrations below. Standard Voice Your browser does not support the audio element. Neural Voice Your browser does not support the audio element. AI-Generated Text-to-Speech AI Assistant’s text-to-speech feature takes voice narration to the next level, using generative AI technology to create highly realistic voices. You can customize the voices to fit your content needs, making the experience feel more personal and engaging to your learners. Here’s an overview of when AI text-to-speech may be the best option for bringing your narrations to life. You need certain (other) languages. AI Assistant allows you to broaden your reach with support for up to 32 languages, depending on the model used to generate narration—including some with multiple accents and dialects. The table below lists 11 languages you can only find in AI text-to-speech. Bulgarian Croatian Filipino Greek Hindi Hungarian Indonesian Malay Slovak Ukrainian Vietnamese You want an easier process. AI Assistant’s intuitive interface helps you quickly generate narration in any supported language. Simply select a voice and enter a script—AI Assistant handles the rest. Even though the voice description may note a specific accent or language, AI Assistant still generates narration in the language used in your script. You need highly customizable voices to create a personalized audio experience. AI Assistant lets you control everything, from adjusting the balance between steadiness and randomness to determining how closely the AI should adhere to the original voice when attempting to replicate it. Click here to learn how to customize AI-generated voices. You want to impress your learners with lifelike, context-aware voices. AI Assistant’s text-to-speech adapts to the tone, emotion, and nuances of your content or script. Here’s an example of a text-to-speech narration created using an AI-generated voice. AI-generated Voice Your browser does not support the audio element. That said, AI text-to-speech has its own drawbacks. For example, because the underlying models don’t support SSML phoneme tags, AI text-to-speech has limited SSML options, as mentioned above. If you have special terminology or pronunciation, indicating that can be harder without full SSML support. And while AI text-to-speech does support the break tag <break time=“1.5s” /> if you want to manually control pacing, note that an excessive number of break tags can potentially cause instability. In addition, AI text-to-speech offers a huge variety of options but no specific guidance on which voices work best for a given language. Finding just the right voice can require a lot of experimentation—that may be time you don’t have. Pro tip: Keeping a reference list of voices that work for specific languages in your courses can help for your next project. Choose What Works For You AI-generated voices clearly have the edge over classic text-to-speech options on voice quality. However, if you (or your organization) are still on the fence about adopting generative AI in your content creation process or have specialized needs, classic text-to-speech is still there to help you create engaging audio interactions. You get to decide what sounds right for your learners—and for your own content and workflow.133Views0likes0Comments
