语音到语音 AI 模型与服务商排行榜

Name: AA-Speech to Speech Index
Creator: Artificial Analysis
License: https://artificialanalysis.ai/docs/legal/Terms-of-Use.pdf

语音到语音模型与 API 服务商的分析和比较。Artificial Analysis 从推理质量、对话动态、生成时间和价格等多个维度，分析了语音到语音模型及其托管服务商。

如需了解更多详情，请参阅方法论页面。

文本转语音排行榜

语音转文本排行榜

Speech Agent Arena 排行榜

Speech Agent Arena 概览

AA-Speech to Speech Index

Weighted average of Speech Reasoning, Agentic Performance, Arena Preference, and Task Success Rate results · Higher is better

速度

Time to first audio (seconds) on Big Bench Audio · Lower is better

每小时输入音频成本

Cost to complete fixed 40-question Big Bench Audio subset based on length of input audio, normalized to hourly basis · Lower is better

Artificial Analysis Speech to Speech Index

Weighted average of Speech Reasoning, Agentic Performance, Arena Preference, and Task Success Rate results · Only models with all data available shown · Higher is better

Equal-weighted score across Speech Reasoning (Big Bench Audio), Agentic Performance (𝜏-Voice), Arena Preference (Arena Score), and Task Success Rate. Models must have results for all four components to receive an index score.

Preference Elo is converted into expected preference against an 800-Elo baseline: Arena Score = 100 / (1 + 10^((800 − frozen Elo) / 400)). The rating is frozen when a model first becomes eligible for publication, so it does not track the live Elo shown on the Arena charts.

语音推理

语音推理（Big Bench Audio）

Speech reasoning: based on the Artificial Analysis Big Bench Audio dataset · Higher is better

Big Bench Audio is an Artificial Analysis benchmark comprising 1,000 audio questions adapted from Big Bench Hard, designed to test the reasoning ability of native audio models. Questions span four categories (250 each): Formal Fallacies (determine whether an argument is deductively valid or invalid), Navigate (determine whether navigation steps return to the starting point), Object Counting (count the number of a specific item class from a list of possessions), and Web of Lies (evaluate the truth value of a Boolean function expressed as a word problem). Models receive an audio input and generate an audio output, which is evaluated as correct or incorrect.

智能体表现

智能体表现（𝜏-Voice）

Proportion of replica customer service scenarios resolved while acting as a customer support agent, based on the 𝜏-Voice benchmark · Higher is better · Only full duplex models

𝜏-Voice measures task completion in real-world replica customer service scenarios. A voice user simulator calls the tested model with a specific domain problem (e.g., requesting a flight change, disputing a retail charge, resolving a telecom issue) across domains inherited from 𝜏-bench and 𝜏²-bench. The tested model is prompted as a customer support agent with access to domain-specific tools and a policy document. Each task has a single valid database end-state; evaluation compares final state against this ground truth. Charts use average performance across 3 trials where available. For more detail on the original benchmark see the 𝜏-Voice paper.

注：以下模型基于 1 次试验：GPT-Live-1 (Astra, medium), OpenAI, GPT-Live-1 (Sol, low), OpenAI, GPT-Realtime-2 (Minimal), OpenAI; 以下模型基于 2 次试验：GPT-Realtime-2.1 High, OpenAI

按领域划分的智能体表现（𝜏-Voice）

Proportion of replica customer service scenarios resolved by domain, based on the 𝜏-Voice benchmark · Higher is better · Only full duplex models

Speech Agent Arena

Arena 偏好 Elo

Preference Elo measures human participants' overall preference after they complete the same scenario with two unidentified Speech to Speech models, such as booking a dental appointment · Only includes models with tool calling · Higher is better

Pairwise preference Elo fitted with Bradley–Terry maximum likelihood, with GPT Realtime 1.5 anchored at 1000.

A Default Cascaded System is a provider’s standard voice-agent pipeline: its orchestration harness connects a speech-to-text model, an LLM, and a text-to-speech model. These entries use the provider’s default stack, rather than a custom pipeline optimized by Artificial Analysis.

Tool registration indicates whether tool definitions are pre-registered or supplied during session setup. As of July 2026:

Model	Harness	STT	LLM	TTS	Tool registration
Deepgram Voice Agent	Deepgram Voice Agent API	Nova-3	GPT-4o Mini	Aura-2	Session-level
ElevenLabs Agents	ElevenLabs Agents	Scribe v2 Realtime	Gemini 2.5 Flash	Eleven Flash v2	Pre-registered
Cartesia Line	Cartesia Line	Ink	Gemini 2.5 Flash	Sonic	Pre-registered schemas; Arena execution
Inworld Realtime	Inworld Realtime API	Inworld STT 1	Gemini 2.5 Flash	Inworld TTS 1.5 Mini	Session-level

任务成功率

Proportion of eligible conversations with the correct final task-completing tool call or calls · Higher is better

Percentage of eligible conversations with the correct final task-completing tool call or calls.

Tool registration indicates whether tool definitions are pre-registered or supplied during session setup. As of July 2026:

Model	Harness	STT	LLM	TTS	Tool registration
Deepgram Voice Agent	Deepgram Voice Agent API	Nova-3	GPT-4o Mini	Aura-2	Session-level
ElevenLabs Agents	ElevenLabs Agents	Scribe v2 Realtime	Gemini 2.5 Flash	Eleven Flash v2	Pre-registered
Cartesia Line	Cartesia Line	Ink	Gemini 2.5 Flash	Sonic	Pre-registered schemas; Arena execution
Inworld Realtime	Inworld Realtime API	Inworld STT 1	Gemini 2.5 Flash	Inworld TTS 1.5 Mini	Session-level

对话动态

对话动态（Full Duplex Bench 子集）

Weighted average of pause handling, turn-taking, interruption handling, and backchannel handling from Full Duplex Bench v1 and v1.5 · Higher is better

Metrics from Full Duplex Bench v1 include Pause Handling (% of samples model correctly does not interrupt) and Turn Taking (% of samples correctly takes the turn in conversation). Metrics from Full Duplex Bench v1.5 include User Interruption Handling (% of samples where the model correctly addresses user's interruption) and Backchannel Handling (% of samples where the model correctly continues when a backchannel such as "yeah" or "alright" is played).

对话动态（Full Duplex Bench 子集）——分类明细

各类别得分，按对话动态得分排序 · 基于 Full Duplex Bench v1 和 v1.5 的子集

API 基准测试

每小时输入音频成本（Big Bench Audio 子集）

Cost to complete fixed 40-question Big Bench Audio subset based on length of input audio, normalized to hourly basis · Lower is better

Cost to complete a fixed 40-question Big Bench Audio subset, normalized to an hourly basis using input audio duration. It incorporates audio input, audio output, text input, text output, and separately exposed reasoning/thinking tokens. It does not include cached-token discounts or tool-call costs.

语音推理（Big Bench Audio）vs. 每小时输入音频成本

Big Bench Audio 得分 vs. 每小时输入音频成本，基于固定的 40 题子集测得 · 语音推理越高、成本越低越好

Most attractive quadrant

首个音频延迟

Average time to first audio (seconds) on Big Bench Audio dataset · Lower is better

Number of seconds required to generate the first token of audio output. A lower value represents a faster generation.

语音推理（Big Bench Audio）vs. 速度

Big Bench Audio 得分 vs. 首个音频延迟 · 语音推理越高、延迟越低越好

Most attractive quadrant

Number of seconds required to generate the first token of audio output. A lower value represents a faster generation.

关键指标摘要与更多信息


Alibaba Cloud	Qwen Audio 3.0 Realtime Plus	69.5	99%	98.4%	54.6%	775	77.8%	1.54	4.42	0.03	0.18
Alibaba Cloud	Qwen3.5 Omni Plus Realtime	-	99%	-	-	859	62.0%	2.64	0.00	0.16	0.82
Alibaba Cloud	Qwen Audio 3.0 Realtime Flash	67.7	96%	96.9%	35.9%	850	81.7%	1.55	4.77	-	-
Alibaba Cloud	Qwen3.5 Omni Flash Realtime	-	59%	-	-	841	29.1%	0.79	0.00	0.16	0.82
Alibaba Cloud	Qwen3 Omni Flash	-	59%	72.7%	-	-	-	4.82	1.77	0.61	1.36
Alibaba Cloud	Qwen3 Omni Realtime	-	57%	-	-	-	-	0.88	2.26	0.65	0.99
Amazon Bedrock	Nova 2.0 Sonic (Mar 2026)	-	88%	-	-	974	57.1%	1.14	4.89	0.27	1.08
Boson AI	Higgs Realtime	-	69%	92.8%	18.6%	-	-	1.47	-	-	-
Cofe AI	FLM-Audio	-	16%	62.0%	-	-	-	-	-	-	-
Deepslate	Deepslate Opal	-	85%	85.7%	17.5%	-	-	0.44	6.48	-	-
Google	Gemini 3.8 Live Extended Thinking (High)	82.6	98%	91.9%	68.6%	990	89.1%	1.35	3.50	-	-
Google	Gemini 3.1 Flash Live High	72.0	97%	74.3%	37.7%	1063	71.8%	2.99	1.75	0.35	1.38
Google	Gemini 3.8 Live	76.0	92%	96.1%	30.1%	1083	93.2%	1.18	0.84	-	-
Google	Gemini 2.5 Flash Native Audio Dialog Thinking	-	91%	-	-	-	-	3.87	-	0.35	1.38
Google	Gemini 3.1 Flash Live Minimal	64.2	71%	72.3%	26.2%	1096	74.6%	0.96	1.50	0.35	1.38
Google	Gemini 2.5 Flash Native Audio Dialog	-	69%	-	-	-	-	0.63	1.42	0.35	1.38
Google	Gemini 2.5 Flash Native Audio Preview (Dec 2025)	-	-	44.0%	22.8%	-	-	-	-	-	-
Google	Gemini 2.5 Flash Native Audio Preview (Sep 2025)	-	-	30.3%	-	-	-	-	-	-	-
Krafton	Raon SpeechChat	-	58%	86.2%	-	-	-	0.04	-	-	-
Kyutai	Moshi	-	4%	61.0%	-	-	-	-	-	-	-
NVIDIA	Nemotron Voicechat	-	27%	52.9%	-	-	-	-	-	-	-
NVIDIA	PersonaPlex	-	19%	91.0%	-	-	-	-	-	-	-
OpenAI	GPT-Realtime-2 (High)	73.6	97%	95.3%	39.8%	932	89.8%	1.14	4.14	1.15	4.61
OpenAI	GPT-Realtime-2.1 High	75.2	96%	95.7%	45.7%	928	91.5%	1.21	10.75	-	-
OpenAI	GPT-Realtime-2 (Medium)	-	93%	95.2%	37.4%	-	-	1.22	3.97	1.15	4.61
OpenAI	GPT-Live-1 (Astra, medium)	83.2	90%	94.9%	74.5%	1048	87.4%	1.34	5.83	-	-
OpenAI	GPT-Live-1 (Sol, low)	80.1	89%	97.3%	59.3%	1053	90.9%	1.24	4.47	-	-
OpenAI	GPT-Realtime-2.1 Minimal	70.7	87%	92.7%	38.0%	937	89.4%	0.97	11.31	-	-
OpenAI	GPT Realtime (Aug '25)	69.1	83%	93.9%	30.4%	974	89.4%	0.98	11.08	1.15	4.61
OpenAI	GPT-Realtime-1.5	70.3	81%	95.7%	38.8%	1000	85.1%	0.81	11.44	1.15	4.61
OpenAI	GPT-Realtime-2.1 Mini High	-	75%	91.7%	29.4%	-	-	4.28	3.45	-	-
OpenAI	GPT-Realtime-2 (Minimal)	63.3	72%	96.1%	30.8%	916	84.7%	1.12	3.07	1.15	4.61
OpenAI	GPT-4o mini Realtime (Dec 2024)	-	69%	-	-	-	-	1.27	5.75	0.36	1.44
OpenAI	GPT Realtime Mini (Oct '25)	56.1	64%	95.7%	15.1%	917	79.6%	0.81	3.04	0.36	1.44
OpenAI	GPT-Realtime-2.1 Mini Minimal	54.4	63%	91.8%	22.5%	840	76.7%	0.85	4.60	-	-
OpenAI	GPT-4o audio chatcompletions	-	54%	-	-	-	-	3.38	0.00	3.60	14.40
OpenAI	GPT-4o Realtime (Dec 2024)	-	-	89.8%	27.9%	-	-	1.49	2.04	1.44	5.76
SpaceXAI	Grok Voice Think Fast 2.0 High	81.3	97%	95.1%	56.5%	1011	94.6%	0.70	4.80	-	-
SpaceXAI	Grok Voice Think Fast 1.0	73.7	97%	77.8%	52.1%	908	80.7%	1.25	3.00	-	-
SpaceXAI	Grok Voice Fast 1.0	-	93%	71.6%	27.4%	-	-	0.78	3.00	3.00	3.00
StepFun	StepAudio 3 Realtime	-	100%	98.9%	-	-	-	8.83	-	-	-
StepFun	Step-Audio R1.1 (Realtime)	-	98%	-	-	-	-	1.53	0.00	0.06	1.69
VITA	Freeze-Omni	-	33%	58.7%	-	-	-	-	-	-	-

常见问题

在已评估的 40 个模型中，StepAudio 3 Realtime 以 99.7% 的语音推理得分在 Big Bench Audio 数据集上领先。Qwen Audio 3.0 Realtime Plus 以 99.2% 位居其后，Qwen3.5 Omni Plus Realtime 为 98.7%。

在已评估的 34 个模型中，StepAudio 3 Realtime 以 98.9% 的对话动态得分在 Full Duplex Bench 数据集上领先。Qwen Audio 3.0 Realtime Plus 以 98.4% 位居其后，GPT-Live-1 (Sol, low) 为 97.3%。

按推理质量排名，顶尖的语音到语音模型为：1. StepAudio 3 Realtime（99.7%）、2. Qwen Audio 3.0 Realtime Plus（99.2%）、3. Qwen3.5 Omni Plus Realtime（98.7%）、4. Gemini 3.8 Live Extended Thinking (High)（97.7%）和5. Step-Audio R1.1 (Realtime)（97.6%）。得分基于 Big Bench Audio 基准测试。

按对话动态排名，顶尖的语音到语音模型为：1. StepAudio 3 Realtime（98.9%）、2. Qwen Audio 3.0 Realtime Plus（98.4%）、3. GPT-Live-1 (Sol, low)（97.3%）、4. Qwen Audio 3.0 Realtime Flash（96.9%）和5. Gemini 3.8 Live（96.1%）。得分基于 Full Duplex Bench 基准测试。

Raon SpeechChat 的首个音频延迟最低，为 0.04秒，其次是 Deepslate Opal（0.44秒）和 Gemini 2.5 Flash Native Audio Dialog（0.63秒）。数值越低，初始响应越快。

Qwen Audio 3.0 Realtime Plus 最实惠，输入音频价格为每小时 US$0.0331，其次是 Step-Audio R1.1 (Realtime)（每小时 US$0.064）和 Qwen3.5 Omni Plus Realtime（每小时 US$0.162）。

Full Duplex Bench 评估语音到语音模型处理真实对话行为的能力：知道何时说话、何时在停顿期间保持沉默、如何回应打断，以及如何识别“嗯”或“对”等反馈语。Artificial Analysis 实现了 Full Duplex Bench v1 和 v1.5 的一个子集。

语音推理（Big Bench Audio）衡量模型能否理解并正确回答以音频形式提出的推理问题。对话动态（Full Duplex Bench）衡量模型能否处理自然的对话流程，包括话轮转换、停顿和打断。一个模型可能擅长其中一项，却不擅长另一项。

实时对话既需要出色的对话动态，也需要较低的延迟。按对话动态得分，StepAudio 3 Realtime（98.9%）、Qwen Audio 3.0 Realtime Plus（98.4%）和 GPT-Live-1 (Sol, low)（97.3%）处于领先。按首个音频延迟，Raon SpeechChat（0.04秒）、Deepslate Opal（0.44秒）和 Gemini 2.5 Flash Native Audio Dialog（0.63秒）速度最快。最佳选择取决于您的使用场景更看重自然的对话流程还是响应速度。

随着新模型和服务商的加入，基准测试会定期更新。性能指标会持续监测，以反映服务商当前的能力和价格变化。

语音到语音 AI 模型与服务商排行榜

相关链接

AA-Speech to Speech Index

速度

每小时输入音频成本

Artificial Analysis Speech to Speech Index

Artificial Analysis Speech to Speech Index

Artificial Analysis Speech to Speech Index

语音推理

语音推理（Big Bench Audio）

Speech Reasoning

智能体表现

智能体表现（𝜏-Voice）

𝜏-Voice

按领域划分的智能体表现（𝜏-Voice）

𝜏-Voice

Speech Agent Arena

Arena 偏好 Elo

Preference Elo

Default Cascaded System

任务成功率

Task Success Rate

Default Cascaded System

对话动态

对话动态（Full Duplex Bench 子集）

Conversational Dynamics

对话动态（Full Duplex Bench 子集）——分类明细

Conversational Dynamics

API 基准测试

每小时输入音频成本（Big Bench Audio 子集）

Cost per Hour of Input Audio

语音推理（Big Bench Audio）vs. 每小时输入音频成本

Cost per Hour of Input Audio

Speech Reasoning

首个音频延迟

Time to First Audio

语音推理（Big Bench Audio）vs. 速度

Time to First Audio

Speech Reasoning

关键指标摘要与更多信息

常见问题

哪个语音到语音模型的推理质量最好？

哪个语音到语音模型的对话动态最好？

顶尖的语音到语音模型有哪些？

按对话动态排名，顶尖的语音到语音模型有哪些？

哪个语音到语音模型最快？

哪个语音到语音模型最便宜？

什么是 Full Duplex Bench（FDB）？

语音推理和对话动态有什么区别？

哪个语音到语音模型最适合实时对话？

语音到语音基准测试多久更新一次？