Yes. All data on the Symbl.ai platform is encrypted, and all connections are secured using 2048 bit AES encryption. Over the wire, data is encrypted using RSA 2048 bit keys. At rest, data is encrypted using AES-256.
We store transcripts, insights, and metadata of conversations, which are persisted on the Symbl.ai back-end. All data access is logged and audited regularly. We maintain logs for a minimum of 3 months, encrypted at rest. By default, we do not store audio recordings. In cases where storing is required, audio recordings are securely stored and encrypted at rest using 2048 bit RSA. For more information, see Security - Symbl.ai.
Zero-shot learning is a new machine learning technique. It helps machine learning environments such as neural networks arrive at correct answers in very few tries. Symbl.ai uses zero-shot learning in natural language processing. We provide algorithms with examples of real-world conversations before using them to process new conversations. For an overview, see Improving AI-Assisted Conversations With Zero-Shot Learning.
In a free forever plan, Streaming API has a limit of 5 concurrent connections, and Async API has a limit of 5 concurrent jobs. After you upgrade your account to the Pro Plan, it has a limit of 50 concurrent connections for Streaming API and 50 concurrent jobs for Async API. If you need concurrency for more than 50 connections, we support higher concurrency in our enterprise plan, you can contact us to discuss your needs.
You can learn more about our plans at Symbl.ai Pricing | Pay As You Go or Volume-Based Discounts.
Conversation intelligence exists on Symbl.ai until you explicitly delete it. When you process a conversation, you receive a
conversationId. The conversation ID and the transcript of the conversation remain available until you delete them using Delete conversation.
Persistence also applies to all Conversation intelligence associated with the conversation, including bookmarks, entities, and trackers.
We do not save or store any source files that you link or upload, including audio, video, and text files. These source files are automatically deleted after processing.
Symbl.ai offers custom rules for data isolation and data retention with a dedicated and customizable deployment, which also comes with enhanced concurrency. For more information about customizing data isolation and retention, contact our sales team.
The only information with a set expiration is your access token, as described in Authenticate.
The Async API used to Process Text supports up to 15,000 words in a single request.
If you want to process larger audio or video files, or text payloads larger than these limits, you can add content to an existing conversation using Append audio (file), Append audio (URL), Append video (file), Append video (URL), or Append text.
Supported Async API audio formats are: mp3, wav, amr, aac, ac3, aiff, flac, ogg, opus, wma, and m4a.
When you Process Audio, we support mono and stereo channel audio files for combined audio of all speakers. We also support processing multi-channel audio files with separate channels for each speaker.
If you have any other file type, you need to convert it to a supported file type. We recommend using ffmpeg or sox depending on your environment.
The only supported Async API video format is mp4. If you have video in another format, you must convert it before submitting to the Async API.
Yes. We support processing multi-channel audio files with separate channels for each speaker using
channelMetadata parameters. For more information, see Submit audio (file) or Submit video (file).
Our Async APIs support multiple popular languages. For the complete list, see Supported Languages.
If the Diarization feature is set to
true, it takes priority over Speaker recognition per Channel.
You can view Async API logs on platform.symbl.ai at API Logs > Async API Logs.
You can view the Streaming API logs on platform.symbl.ai at API Logs > Streaming API Logs.
The platform experience enables you to process conversation data, test APIs, and explore Symbl.ai Conversation Intelligence features without writing a single line of code. Log in to your account at platform.symbl.ai to get started.
On platform.symbl.ai go to Playground to easily process conversations in Streaming, Async, and Telephony formats. You can complete the end-to-end process in the playground and generate Conversation Intelligence without writing any code. In Streaming and Async conversations, you can use sample content or provide your own.
You can also save your results, which appear in Playground > Saved sessions. This is helpful if you process a conversation and want to save the insights, or apply different features at a later time.
The API Explorer is a REST API interface that gives you the ability to test and explore APIs with a just few clicks. Simply log in to platform.symbl.ai and click API Explorer, then complete round-trip API requests and responses directly in the platform.
From the API Explorer you can test drive many Symbl.ai features including Trackers, Topics, Analytics, Transcription, Follow-Ups, Questions, Entity Detection, Action Items, and Summary.
Audio streaming using microphone, sample audio files that simulate streaming and uploading, sample text files, and connecting by phone.
Use the Shareable Link button from the Playground dashboard to get a link you can share with anyone.
For a list of languages supported by pre-recorded audio files, see Supported Languages.
English is the only supported language for streaming and voice calls.
Yes, the conversations you process through the Conversation Intelligence tool are counted against your monthly balance.
You can complete the current conversation. You won’t be able to complete a new conversation until the next month, or you upgrade your account.
We recommend using a phone number you have immediate access to, such as a personal phone number or a Zoom conference line.
We will be expanding support for voice calls into other countries in the future. Please contact [email protected] to let us know if there is a country or region you are interested in having supported.
Yes, DTMF digits are supported.
Yes. For details, see Zoom - Symbl.ai Meeting App.
Yes, dialing into a Zoom conference line is supported with the voice call feature. It will capture audio from all participants, but the conversation mode tool doesn’t yet support specifying how many participants (speakers) are interacting during a call. This will result in all transcriptions appearing as coming from a single speaker.
No, you can share the link with as many people as you want.
The link expires 24 hours after it is generated
No, they can freely use the link until it expires. If more time is needed, you can simply go back to the dashboard and click the create shareable link button again to get a new link, and then share it.
Signed URLs cannot be revoked, so the person would have access until the URL expires. However, you can delete the session that was shared, and create a new session using the same audio or text.
Can I use an app ID and app secret from another account to make requests through the API Playground?
No, the app ID and app secret are read-only, and can’t be changed from within Playground. You can API Reference docs or the Postman collection to experiment and test Symbl.ai APIs
There is no limit for Pro and Enterprise users. Free users can stream up to their monthly limit. For current plans and details, see Pricing.
Don’t worry! Audio streams will disconnect automatically after one minute of inactivity.
Yes, use the Voice Call option to dial in to an existing US or toll-free conference line by entering the phone number and any DTMF digits.
Yes, the sample audio clips are being streamed live. All data appearing in real-time is based on what is happening over that audio stream, so none of what you are seeing is pre-rendered.
The audio script provides a simple way for speakers to see Symbl.ai features while streaming audio from your device. Speakers are encouraged to say whatever they want to see how Symbl.ai processes the conversation.
No, not at this time.
The tool is free to use for as long as you want.
Yes, absolutely, that’s what it is here for!
Sentiment, Questions, Follow-Ups, Topics, and Action Items.
Our pricing plans let you start free, pay as you go, and receive discounts as you scale. We offer three levels of pricing plans:
- Developer - Does not require a credit card and can be free forever within usage limits.
- Pro - Is a pay-as-you-go plan with no usage commitment required.
- Enterprise - Features customizable high-volume usage plans.
For current plans and details, see Pricing.
Symbl.ai billing is handled through platform.symbl.ai. Log in to your account and from the left navigation panel go to Billing. This page shows your account level, current balance, and usage to date.
If you want to add or upgrade your account, click Upgrade Now to enter billing address and payment information.
To remove an active credit card from your account, contact [email protected]. You are responsible for any outstanding charges on the account.
The Bookmarks feature empowers customers to capture relevant moments in a conversation, as it happens. Bookmarks enable customers to capture, label, and summarize key moments in real-time and asynchronous conversations.
You can use Bookmarks to:
- Capture key moments in a conversation to reference later or share with others.
- Label parts of conversation for easy navigation.
- Get a summary of a single or multiple bookmarks that correspond to relevant parts of a conversation.
Bookmarks enable customers to influence what is considered important to them for each conversation, in real time or from recorded files, and to use that influence to bring insightful intelligence and efficiency to their daily operations.
Use Bookmarks to efficiently point directly to the moments you determine are most relevant, saving time and energy sharing and listening to the bookmarked content. Also, a single bookmark or all bookmarks from a conversation can be summarized to provide an informative summary of the most important moments to store for referencing later or for sharing with others.
The Symbl.ai Bookmarks feature complements its other conversation intelligence features by enabling customers to take action to capture relevant moments in addition to the relevant topics, action items, and trackers that occur in the conversation.
For example a sales representative might have a tracker set up to track mentions of some known competitors but during the call the rep also bookmarked a customer pain point that led to the mention of a competitor. The rep can then easily share the tracked competitor and the moment that was worth capturing that led to it with the customer pain point. Bookmarks also leverage the Summary feature to create a succinct summary of single or multiple bookmarks.
When you first implement Bookmarks, test how the default duration works for your conversations. Symbl.ai provides a default duration to ensure out-of-the-box value. Some customers conversations require adjustments to the duration captured for each bookmark.
It also helps to organize Bookmark groups with standard labels. Consistent labeling improves storage and reference summarization when dealing with multiple bookmarks.
The Summary feature provides a compact version of Bookmarks. You can get a Summary of a single or multiple Bookmarks that correspond to relevant parts of a conversation.
The Entity Detection feature identifies sensitive information in both recorded and real time conversations. Some detected entities include people’s names, organizations, locations, medical codes, email address, SSN, banking information, driver’s license, credit card information, phone number, people’s age, occupation, date of birth, passport number, and money amounts. This feature detects and returns the entities detected in the API response, along with the locations of those entities within the conversation transcript.
Entity Detection helps you identify the occurrence of valuable entities from your conversations and results can be obtained with a simple API request. Companies can monitor social media trends, and extract entities from social media forums like Twitter and Reddit. For example, you can use Entity Detection to identify which stocks are being discussed by users that might cause a stock market crash.
Use entities to tag conversations, documents, and use tags to efficiently search through massive volumes of conversations and documents. You can identify entities from conversations to automatically fill out forms such as, fill out customer name, address, phone number, and appointment date and time from a sales call. You can also create custom entities and detect them in the conversation, both asynchronously and in real-time.
Entity Detection helps users identify PII, PCI, and PHI entities out-of-the-box. You can also use the Symbl.ai Trackers feature to find similar word and phrases, and add all the entities to be identified. In an upcoming release, you will be able to identify entities in insights, questions, action items, follow ups, and in all Conversation Intelligence provided by Symbl.ai.
If you are new to Symbl.ai, login to the Symbl.ai Platform and use the API Explorer to detect entities. You can use the sample conversations provided or upload your own conversations and process to identify entities. Entity Detection is enabled by default when you process a conversation. Managed entities are created and managed by Symbl.ai, so they are well researched, benchmarked against thousands of conversations, and are continuously improved to save you time and effort.
You can choose from managed entities curated by Symbl.ai, or create your own entities using the Create custom entities request. To create a custom entity, you need to provide the entity’s type, subtype, category, and values:
- Type is the class in which you would like this entity included. Multiple entities can have the same type.
- Subtype is the name identifying the new entity.
- Category should always be
- Values are the entities that have to be identified as a part of the new entity.
For example, if you want to identify Tesla vehicles as an entity:
- Type =
- Subtype =
- Category =
- Values =
When you create a custom entity, Symbl.ai generates a unique entity ID. You can update custom entities by entity name and entity ID. For more information, see the Entity Detection docs.
Will I be compliant with industry standards like PCI SSC, HIPAA, GDPR, and CCPA, if I use Entity Detection?
Symbl.ai Entity Detection currently identifies entities for Payment Card Industry (PCI) Security Standards Council (SSC) compliance. Entity Detection can help you to be partially compliant for Health Insurance Portability and Accountability Act (HIPAA), General Data Protection Regulation (GDPR), and California Consumer Privacy Act (CCPA).
Yes, both custom entities and out-of-the-box entities can be detected in real-time. Entities can be detected in real-time using the Streaming API, in both audio and video formats.
No, Symbl,ai charges by conversation processing times for the calls sent to Async API and Streaming API. Entity Detection does not incur additional charges beyond the processing of the original conversation.
Symbl.ai detects Personal Identifiable Information (PII), Payment Card Industry (PCI), and Protected Health Information (PHI) entities out-of-the-box. Current supported countries are Australia, Canada, India, United Kingdom, and United States.
For a complete list of entities, see Entity Detection.
Trackers is a Symbl.ai feature that enables customers to track occurrences of similar keywords and phrases in conversations. Managed Trackers is a library of trackers that Symbl.ai has curated and tested for various categories of conversations such as Sales, Contact Center, and Recruitment. The library also includes general trackers that apply across all categories
You can use Managed Trackers as delivered, or customize them by including specific keywords and phrases you want to track. It is easy to Get Started with Trackers. Log in to platform.symbl.ai and use Trackers Management > Managed Trackers Library to add to Your Trackers.
Managed Trackers help you get valuable insights, right out of the box, as you process your first conversation. Managed Trackers are curated by Symbl.ai after analyzing thousands of conversations and are benchmarked for accuracy and precision for use-case specific conversations.
Now, you do not need to invest time and energy defining a tracker and its vocabulary, testing the tracker on multiple conversations, and fine-tuning the tracker multiple times based on results. The Managed Trackers Library takes away all that work and lets you simply choose the trackers you want to use for your application thus reducing the time to go to market. Symbl.ai is continuously testing and improving Managed Trackers, so you do not have to spend time improving your trackers as you process new conversations.
Recommendations are suggestions for trackers to your library of trackers. At any point in time, recommendations are based on data collected from conversations that you have processed in the last 30 days, in real time or asynchronously.
Use Tracker Recommendations to discover new trackers that you can add to Your Trackers. Recommendations are trackers that are not currently in Your Trackers, but include phrases or intentions that Symbl.ai detects in the conversations that you recently processed.
Trackers is available for real-time conversations processed via Streaming API and recorded conversations processed via Async API. As part of this launch, we are launching trackers in GA, adding the managed library and tracker recommendations. The new features are designed to solve the cold start problem that customers face today when they first use trackers. Customers can now easily get started with Trackers by selecting few or all the managed trackers offered by Symbl.ai. Recommendations get generated as soon as customers process their first conversation with Symbl.ai. Tracker detection is now enabled by default for all conversations and can be disabled for a conversation if required.
Choose Managed Trackers to get started. If you are new to Symbl.ai or the Trackers feature, start with selecting trackers from the Managed Trackers Library. Managed trackers are created and managed by Symbl.ai. They are well researched, benchmarked on thousands of conversations, and are continuously improved to save you time and effort.
When you see Recommendations while creating a custom tracker, choose from recommended trackers when possible. Recommendations are generated from the Managed Trackers Library. For the reasons previously mentioned, it is best to choose managed trackers when possible.
Periodically check tracker recommendations. We continue to improve the Managed Trackers library which will include new trackers and refining existing Managed Trackers. Periodically check tracker recommendations on the platform to identify new trackers that are relevant to the conversations that you are processing.
Test Trackers using API Explorer. When you add Managed Trackers or create custom trackers, we recommend that you use the API explorer on the platform to test if the new trackers are getting detected in conversations as expected. If results are not as expected, customize the trackers by adding vocabularies as needed.
Choose phrases over keywords. When creating custom trackers or customizing managed trackers, we recommend that you write complete phrases as vocabularies rather than single word vocabularies. Single words often lead to false positives in the tracker detection results.
You can use most punctuation in Trackers phrases. The following special characters are not allowed.
|[ ]||Square Brackets|
If the system encounters one of these special characters, it returns the error message:
Special characters are not allowed in vocabulary. Please refer to Symbl docs for more information.
You can use the Management API Create tracker operation. You can also use the platform.symbl.ai Trackers Management > Your Trackers. You can use both methods to create, view, edit, and delete Trackers.
We recommend adding at least 5 vocabulary terms and a maximum of 50 per Tracker.
You can create up to 700 Trackers per account. This is the upper limit and cannot be increased.
Updated 8 months ago