SiriusXM has agreed to provide accessible podcast transcripts for deaf and hard-of-hearing listeners under a legal settlement that resolves a lawsuit filed in 2021. The requirements apply to podcasts offered through both SiriusXM and Pandora, and the agreement goes substantially further than simply converting spoken words into a block of automated text.
The settlement calls for transcripts that identify speakers, use punctuation and paragraph breaks, and describe relevant non-speech audio, including atmosphere and sound effects. It also requires human review when speech-recognition technology assists in producing the text. SiriusXM must additionally create podcast-specific dictionaries, maintain a style guide developed with input from the plaintiffs, and train staff on Deaf culture and communication access.
What the settlement requires
At its center, the agreement recognizes that access to an audio-first program depends on more than a rough record of dialogue. A useful transcript needs to convey who is talking, where one exchange ends and another begins, and when meaningful audio changes the context of what is being said.
- Speaker identification: Transcript readers should be able to tell which host, guest, narrator or participant is speaking.
- Relevant sound descriptions: Atmospherics and sound effects that matter to the program must be represented in text.
- Readable formatting: Proper punctuation and paragraph breaks are specifically required rather than treated as optional polish.
- Human cleanup: Where automated speech recognition is used, a person must review the result.
- Podcast-specific terminology: SiriusXM must build custom dictionaries of words and phrases for individual shows.
- Shared editorial standards: An active style guide will be developed with plaintiffs’ input.
- Staff training: The company must train personnel on Deaf culture and communication access.
Those details matter because spoken-word programming is rarely as simple as one person speaking clearly into a microphone. Podcasts can feature overlapping participants, recurring names, uncommon terms, accents, music, environmental audio, recorded clips and intentional pauses. Without careful editing and context, an automated transcript can preserve some words while losing the meaning of a conversation.
Why speaker labels and sound cues are not minor extras
Speaker identification means attaching the right identity to a line of dialogue. In an interview show, this lets a reader distinguish the host’s question from a guest’s answer. In a roundtable, it helps establish whether an apparent disagreement is coming from one participant or another. In narrative programming, it can distinguish narration from archival recordings or quoted material.
Atmospherics are the background sounds that establish an audio setting or mood. They may be as simple as a crowd, traffic, rain, applause, or the ambience of a room. A sound effect can also signal a scene transition, a joke, a reveal or an action that is not described aloud. The settlement does not suggest that every bit of incidental audio needs exhaustive transcription. Its focus is on audio that is relevant to understanding and enjoying the work.
That distinction is important. Accessibility is not achieved by reproducing noise for its own sake; it is achieved when a listener who cannot access the audio is given the information needed to follow the program on comparable terms. A bracketed description of meaningful laughter, music or a scene-setting sound can change how an exchange reads. Conversely, a transcript that provides dialogue alone can make a carefully produced show feel incomplete.
Paragraphing and punctuation serve a similar function. Raw speech does not arrive with commas, sentence endings or neatly separated topics. Transcript editors must make decisions that help readers navigate the material while preserving what was said. In practical terms, well-structured text makes it easier to scan an interview, return to a point, read a long discussion without fatigue, and understand interruptions or shifts in tone.
Why human review is a major part of the agreement
The settlement permits speech-recognition tools to assist with transcription, but it requires an actual person to review output. That is a consequential safeguard. Speech recognition is a technology that turns recorded speech into text by matching audio patterns to likely words and phrases. It can accelerate the creation of a draft, but it can also mishear names, specialized language and similar-sounding words.
Podcast-specific data dictionaries are designed to address a portion of that problem. A recurring show may have distinctive terminology, regular contributors, proper names, segment titles, technical jargon or community references. Supplying a custom dictionary gives a transcription process a better basis for rendering those terms accurately. But dictionaries alone cannot resolve whether a word was used sarcastically, whether two people spoke simultaneously, or whether a sound conveys essential context. Those are among the reasons editorial review remains necessary.
The agreement’s style-guide requirement also points to consistency as an accessibility issue. A style guide is a living set of editorial rules: how speakers are labeled, how non-speech audio is described, how uncertain words are handled, and how formatting is applied. Input from the plaintiffs gives the people affected by the access gap a role in shaping those standards, instead of leaving all of those decisions to a platform’s internal workflow.
A case that began in 2021
The original lawsuit was filed in 2021. At that time, SiriusXM and Pandora did not offer podcast transcripts for the deaf and hard-of-hearing community. The suit characterized that lack of access as a violation of the Americans with Disabilities Act.
The Americans with Disabilities Act, commonly shortened to the ADA, is a U.S. civil-rights law intended to prohibit disability-based discrimination in covered areas of public life. In this matter, the dispute concerned whether people who are deaf or hard of hearing could meaningfully access podcast programming when the platforms offered it only in audio form.
The settlement resolves that dispute with concrete operational obligations rather than a vague statement of intent. It specifies qualities the transcripts must have, requires review beyond automated generation, and addresses the internal practices that influence quality over time. That structure is why the agreement could matter beyond a single catalog of shows: it creates a detailed model for what accessible transcription can involve.
What it could mean for podcast audiences
For deaf and hard-of-hearing people, transcripts can make a program accessible that otherwise would not be. They also permit a different way of engaging with a podcast: reading at one’s own pace, jumping to a particular answer or topic, checking an unfamiliar name, or revisiting a point without replaying audio. The settlement is focused on communication access, but the formatting and accuracy requirements are central to whether that access is genuinely useful.
Plaintiff James Munro framed the issue as equal participation in a major form of communication: “Podcasts have become a huge part of how we share news, ideas, stories and information, and deaf people deserve equal access to all of it.” Munro expressed hope that the agreement would help make inaccessible podcasts a thing of the past.
That broader point is especially notable because podcasts cover a wide range of subjects and formats. They can carry interviews, commentary, narrative reporting, comedy, fan discussion and cultural analysis. If the only version of a program is audio, a person unable to hear it is excluded not only from the show itself but potentially from the discussions it starts. The agreement treats a transcript as part of access to the work, not merely a searchable bonus feature.
A potential benchmark, not an industry-wide solution
The settlement applies to SiriusXM and Pandora podcasts; it does not automatically impose the same obligations on every podcast service. Other platforms still may not have robust transcription systems. That limitation is worth keeping clear. The agreement is a significant development for the services it covers, but it is not proof that podcast accessibility has been solved across the industry.
Nick Kiego, chief executive officer of the National Association of the Deaf, said he hopes the result becomes a benchmark for other providers of podcast entertainment. His emphasis was on opening access to conversations ranging from politics to pop culture.
As an industry example, the settlement establishes several practical reference points for what “transcripts available” should mean. A platform can technically publish text while still delivering a poor experience if speaker names are missing, sounds that alter context are ignored, specialized vocabulary is repeatedly wrong, or no editor checks machine-generated errors. The requirements here draw attention to the difference between transcript presence and transcript quality.
That is also relevant to the wider entertainment business. Audio programming increasingly intersects with film, television, music, sports, games and celebrity culture, while audiences often encounter clips and discussions across multiple platforms. Accessibility practices determine who can take part in that flow of information and conversation. For more entertainment-industry context, see our coverage of how creative formats shape audiences and access.
The practical test will be quality over time
The settlement lays out a demanding approach, but its value will ultimately depend on implementation. The most important features are not limited to the first publication of a transcript. Podcast catalogs evolve, new shows introduce new voices and vocabulary, and editorial standards must be maintained as production routines change. The custom dictionaries, active style guide and staff training requirements are all aimed at that continuing work.
For readers, the clearest signs of success would be transcripts that are easy to find, reliably identify speakers, preserve meaningful audio context and read like carefully prepared text rather than uncorrected software output. For creators and distributors, the agreement underscores that access is a production and editorial responsibility, not simply a button that can be added after publication.
SiriusXM’s commitment closes a case that had been pending since 2021 while setting detailed expectations for its SiriusXM and Pandora podcast offerings. The agreement’s strongest contribution may be its specificity: it describes accessibility as accurate, contextual, reviewed communication—and gives that standard an operational form.





