Search “Speech”

Perplexity Mac and web voice mode switch to gpt-live-1

Perplexity announced that the voice mode of its Mac app and web version now uses OpenAI's gpt-live-1 model. According to Testing Catalog, users can also run multiple sessions on Perplexity Computer and manage tasks and Projects; the poster said voice is their favorite interaction interface. 9to5Mac reported that this upgrade enables the Mac app's voice mode to simultaneously handle multiple Computer sessions, launch new tasks, and create and manage Projects within a single voice session. There are currently no further developments.

Perplexity Mac and web voice mode switch to OpenAI's gpt-live-1

Perplexity announced that the voice mode of its Mac app and web version now uses OpenAI's gpt-live-1 model. According to Testing Catalog, users can also run multiple sessions on Perplexity Computer and manage tasks and Projects; the poster said voice is their favorite interaction interface. 9to5Mac reported that this upgrade enables the Mac app's voice mode to simultaneously handle multiple Computer sessions, launch new tasks, and create and manage Projects within a single voice session. There are currently no further developments.

View story:Perplexity Mac and web voice mode switch to gpt-live-1

Artificial Analysis ranks Phonic 1 first on the Speech to Speech Index

Artificial Analysis's evaluation shows that Phonic's Phonic 1 (High) ranks first among 22 models on the Speech to Speech Index with a score of 87.1, narrowly ahead of xAI's Grok Voice 3.0 (86.9 points). In the same evaluation, Phonic 1 ranks first on the Tau Voice agent tasks with 79.9%, leading the second place by 5.4 percentage points; on Big Bench Audio it ranks third with 98.9%, with a first-audio latency of 0.61 seconds. Phonic 1 is available through the Phonic API, with input audio costing $0.07 per minute and $4.20 per hour. Another report states that Phonic is a 10-person startup, which claims that Phonic-1 can execute queries and bookings before the caller has finished speaking; the report also mentions that Phonic claims a score of 90.2 on the Cekura call test set, while the public Cekura leaderboard lists Phonic v1 at 68.3%.

Artificial Analysis ranks Phonic 1 first on the Speech-to-Speech Index

Artificial Analysis's evaluation shows that Phonic's Phonic 1 (High) ranks first among 22 models on the Speech to Speech Index with a score of 87.1, narrowly ahead of xAI's Grok Voice 3.0 (86.9 points). In the same evaluation, Phonic 1 ranks first on the Tau Voice agent tasks with 79.9%, leading the second place by 5.4 percentage points; on Big Bench Audio it ranks third with 98.9%, with a first-audio latency of 0.61 seconds. Phonic 1 is available through the Phonic API, with input audio costing $0.07 per minute and $4.20 per hour. Another report states that Phonic is a 10-person startup, which claims that Phonic-1 can execute queries and bookings before the caller has finished speaking; the report also mentions that Phonic claims a score of 90.2 on the Cekura call test set, while the public Cekura leaderboard lists Phonic v1 at 68.3%.

View story:Artificial Analysis ranks Phonic 1 first on the Speech to Speech Index

NVIDIA Releases Open-Weight Speech Transcription Model Nemotron 3.5 Transcribe

Alex Heath says Granola's new voice input eliminates the need for a separate dictation app

Granola has launched a dictation feature: users can simply press the Fn key on their Mac and speak, and the tool automatically cleans up and formats the speech and outputs the final text, which can be pasted and used in all apps; the feature is free with no word limit, and iOS and Windows versions are coming soon. Alex Heath said he often uses Granola to handle meetings, and now he no longer needs to install a separate dictation app, and said he has used it for voice input on his Mac for a whole day, with a smooth experience.

Alex Heath says that with Granola's new voice input, there's no need to install a separate dictation app

Granola has launched a dictation feature: users can simply press the Fn key on their Mac and speak, and the tool automatically cleans up and formats the speech and outputs the final text, which can be pasted and used in all apps; the feature is free with no word limit, and iOS and Windows versions are coming soon. Alex Heath said he often uses Granola to handle meetings, and now he no longer needs to install a separate dictation app, and said he has used it for voice input on his Mac for a whole day, with a smooth experience.

View story:Alex Heath says Granola's new voice input eliminates the need for a separate dictation app

Aravind Srinivas announces new voice mode for Pro and Max users

Perplexity CEO Aravind Srinivas announced that the new voice mode is available to all Pro and Max users, covering the web and the Mac app. This mode lets users talk to their computer by voice and advance multiple workflows in parallel. There are currently no subsequent developments or facts.

Aravind Srinivas announces new voice mode for Pro and Max users

Perplexity CEO Aravind Srinivas announced that the new voice mode is available to all Pro and Max users, covering the web and the Mac app. This mode lets users talk to their computer by voice and advance multiple workflows in parallel. There are currently no subsequent developments or facts.

View story:Aravind Srinivas announces new voice mode for Pro and Max users

Aravind Srinivas says voice is the natural form of using a computer

Perplexity's Houssein Djirdeh announced that Perplexity's web version has launched a new voice mode, supporting natural conversation, full voice operation of the app, and multiple parallel sessions, based on GPT-Live-1. Perplexity CEO Aravind Srinivas reposted the news and commented that voice is the natural form of using a computer. There is currently no further development.

Aravind Srinivas proposes that voice is the natural form of using a computer

Perplexity's Houssein Djirdeh announced that Perplexity's web version has launched a new voice mode, supporting natural conversation, full voice operation of the app, and multiple parallel sessions, based on GPT-Live-1. Perplexity CEO Aravind Srinivas reposted the news and commented that voice is the natural form of using a computer. There is currently no further development.

View story:Aravind Srinivas says voice is the natural form of using a computer

Gnani AI releases speech recognition model Prisma v3.0

Gnani AI has released Prisma v3.0, an automatic speech recognition model that supports nine Indian languages, including Hindi and Bengali, as well as English, and can transcribe multilingual conversations with intra-sentence Hindi-English code-mixing. According to Analytics India Magazine, the model achieves a macro-average WER of 16.93 on the Vistaar benchmark, lower than ElevenLabs Scribe v2's 19.98, Microsoft Azure Speech-to-Text's 20.51, and Deepgram Nova 3's 26.57. It offers four transcription modes—Verbatim, Transcribe, Code-mix, and Transliterate—and supports up to 200 custom vocabulary terms per request, which can raise the recall rate for these terms from 70.9% to 90.3%. No further updates have been disclosed at this time.

Rivian discloses expansion of its joint venture with Volkswagen and licensing of AI voice assistant technology

In its Q3 2026 earnings report, Rivian disclosed that it has licensed its AI Voice Assistant technology to the Volkswagen Group through the Rivian and VW Group Technology joint venture, as part of the two companies' electrical architecture and software collaboration. TechCrunch AI reported that Rivian is licensing some elements of its AI voice assistant technology to the Volkswagen Group as part of their $5.8 billion joint venture; the assistant launched on the R1S and R1T in May 2026 and on the R2 in October, and may first appear in Volkswagen's low-cost ID EVERY1 electric vehicle. Rivian did not disclose the licensing fee or specific models, saying only that this demonstrates the scalability of its in-house platform. Unite.AI reported that Rivian's software and services revenue for the quarter was $605 million, up 45% year over year, of which $354 million (59%) came from the joint venture; Rivian Assistant launched on the R2 in October 2026 and gained Gmail integration through an OTA update.

Neuralink publishes early feasibility study of VOICE speech restoration project

On October 28, Neuralink published a blog post disclosing the early feasibility study of its speech restoration project VOICE. The blog post revealed that an ALS participant reached an output speed of 75 words per minute in the first week of use, with a word error rate of 5.8% on a large vocabulary of 125,000 words. The study's principal investigator, Dr. Nader Pouratian, has presented the relevant data at the 2026 NEALS annual meeting. The study plans to enroll approximately 6 participants who have lost the ability to communicate due to ALS, spinal cord injury, or stroke, with the primary endpoint being device- or procedure-related adverse events within 12 months after implantation. Neuralink has not yet published a complete peer-reviewed paper; the above figures come from a single participant, and the company has not explained the types of test tasks, the separate performance of text versus speech output, or long-term effects.

Neuralink discloses early feasibility study of VOICE speech restoration project

On October 28, Neuralink published a blog post disclosing the early feasibility study of its speech restoration project VOICE. The blog post revealed that an ALS participant reached an output speed of 75 words per minute in the first week of use, with a word error rate of 5.8% on a large vocabulary of 125,000 words. The study's principal investigator, Dr. Nader Pouratian, has presented the relevant data at the 2026 NEALS annual meeting. The study plans to enroll approximately 6 participants who have lost the ability to communicate due to ALS, spinal cord injury, or stroke, with the primary endpoint being device- or procedure-related adverse events within 12 months after implantation. Neuralink has not yet published a complete peer-reviewed paper; the above figures come from a single participant, and the company has not explained the types of test tasks, the separate performance of text versus speech output, or long-term effects.

View story:Neuralink publishes early feasibility study of VOICE speech restoration project

DingTalk A1 Doukou Doctor Edition combines voice recording with medical evidence-based capabilities

On October 30, 2026, DingTalk A1 Doukou Doctor Edition was released, combining A1's voice recording and AI meeting notes capabilities with Doukou's medical evidence-based capabilities. According to a China.com Technology report, after doctors finish organizing case discussions or lecture content, they can also query literature, guidelines, and expert consensus around unresolved medical questions. A gynecology department director at a tertiary hospital in Hangzhou reported that on surgery days she can first record special situations and judgments, then organize them after surgery, while for complex outpatient visits she uses Doukou Doctor to look up literature and guideline links. The report states that at this stage it is more like a portable professional assistant capable of recording, distilling, and searching literature than a reconstruction of hospital IT systems.

NVIDIA releases open-weight speech transcription model Nemotron 3.5 Transcribe

View story:NVIDIA Releases Open-Weight Speech Transcription Model Nemotron 3.5 Transcribe

AssemblyAI launches its first text-to-speech model, AssemblyAI-TTS-1

On October 29, 2026, AssemblyAI launched its first text-to-speech model, AssemblyAI-TTS-1, and made it generally available, targeting real-time telephony voice agents, priced at $15 per 1 million characters, with no annual contract and no concurrency tiers. According to a MarkTechPost report, the model is offered only as a hosted API in us-east-1 and eu-north-1, supports HTTP, WebSocket and OpenAI-compatible interfaces, and includes more than 4,000 voices, instant voice cloning and 6 languages; vendor self-testing shows 93.7% entity accuracy and a median time to first byte of 47 milliseconds, but Cekura's independent benchmark measured a streaming latency of 159 milliseconds and a reading accuracy of 81.5%, and at launch it did not yet support SSML, speaking-rate control or emotion tags. On the same day, Artificial Analysis' Provider Voice Arena evaluation showed AssemblyAI-TTS-1 ranked 12th with an Elo of 1227, ranked 16th with a pronunciation robustness score of 81.7%, and generated at 162 characters per second. Later that day, OpenRouter announced that AssemblyAI-TTS-1 was live, targeting voice agent scenarios, with service available from US or EU nodes.

View story:AssemblyAI launches its first text-to-speech model, AssemblyAI-TTS-1

Gnani AI releases speech recognition model Prisma v3.0

Gnani AI has released Prisma v3.0, an automatic speech recognition model that supports nine Indian languages, including Hindi and Bengali, as well as English, and can transcribe multilingual conversations with intra-sentence Hindi-English code-mixing. According to Analytics India Magazine, the model achieves a macro-average WER of 16.93 on the Vistaar benchmark, lower than ElevenLabs Scribe v2's 19.98, Microsoft Azure Speech-to-Text's 20.51, and Deepgram Nova 3's 26.57. It offers four transcription modes—Verbatim, Transcribe, Code-mix, and Transliterate—and supports up to 200 custom vocabulary terms per request, which can raise the recall rate for these terms from 70.9% to 90.3%. No further updates have been disclosed at this time.

View story:Gnani AI releases speech recognition model Prisma v3.0

Rivian discloses expansion of joint venture with Volkswagen and licensing of AI voice assistant technology

In its Q3 2026 earnings report, Rivian disclosed that it has licensed its AI Voice Assistant technology to the Volkswagen Group through the Rivian and VW Group Technology joint venture, as part of the two companies' electrical architecture and software collaboration. TechCrunch AI reported that Rivian is licensing some elements of its AI voice assistant technology to the Volkswagen Group as part of their $5.8 billion joint venture; the assistant launched on the R1S and R1T in May 2026 and on the R2 in October, and may first appear in Volkswagen's low-cost ID EVERY1 electric vehicle. Rivian did not disclose the licensing fee or specific models, saying only that this demonstrates the scalability of its in-house platform. Unite.AI reported that Rivian's software and services revenue for the quarter was $605 million, up 45% year over year, of which $354 million (59%) came from the joint venture; Rivian Assistant launched on the R2 in October 2026 and gained Gmail integration through an OTA update.

View story:Rivian discloses expansion of its joint venture with Volkswagen and licensing of AI voice assistant technology