License & Deployment Mix: 12 tools – 5 OSS, 7 SaaS. (OSS and SaaS counts can overlap when an open-source tool also offers a hosted edition.)
What Is Text-to-Speech Software?
Text-to-speech (TTS) synthesizes spoken audio from written text. It powers voiceovers and narration, IVR and phone-tree prompts, notifications and alerts, audiobook and e-learning content, and the output side of voice assistants and accessibility (screen readers).
As with speech-to-text, the axis that matters most is where the text goes:
- Local / self-hosted engines synthesize on your own hardware; text and the resulting audio never leave the machine or network. Examples: Piper, Coqui TTS, eSpeak NG, Kokoro, Bark.
- Cloud services send text to a vendor API over the internet and stream back audio. They offer the most natural, expressive voices and features like voice cloning, but text leaves your environment and usage is metered. Examples: ElevenLabs, Amazon Polly, Google Cloud Text-to-Speech, Azure AI Speech, Murf, Play.ht, Resemble AI.
Core Concepts
| Concept | Description |
|---|---|
| Voice / speaker | The synthetic voice; may be preset or custom |
| Naturalness (MOS) | Mean opinion score – how human the voice sounds |
| Neural vs formant | Neural voices sound natural; formant (eSpeak) is robotic but tiny |
| Voice cloning | Reproducing a specific voice from samples (consent matters) |
| SSML | Markup to control pronunciation, pauses, emphasis, prosody |
| Streaming | Emitting audio as it synthesizes, for low latency |
| On-device | Runs fully offline, no network call |
The information on these pages was researched by a combination of human review and large language models. To suggest an addition or correction, please contact us. Prepared by Rhodium Systems Inc., author of the ResorsIT platform — a unified IT operations management platform for IT teams and MSPs that integrates a curated suite of open-source, commercial, and SaaS applications into a single system with shared identity, single sign-on, access control, and a common audit trail. Use this catalogue only as a starting point for your own research, and review any tool carefully against your own requirements before relying on it. Catalogue data version 2026.197.
Comparison
Text-to-speech engines serve accessibility, IVR / phone-tree prompts, spoken notifications, and narration. The guiding question is where the text goes: local engines keep it in the customer environment; cloud services send it to a vendor over the internet.
Local vs Cloud (Where the Text Goes)
This is the table that matters most: whether a tool runs entirely on your own hardware or needs an internet connection to a vendor service.
| Tool | Runs | Needs internet? | Notes |
|---|---|---|---|
| Piper | Local | No | Fast neural TTS; fully offline, even on a Raspberry Pi |
| Coqui TTS | Local | No | Offline; many models incl. XTTS voice cloning (GPU helps) |
| eSpeak NG | Local | No | Tiny formant synth; fully offline, robotic voice |
| Kokoro | Local | No | Small high-quality neural model; offline (ONNX) |
| Bark | Local | No | Generative; offline but wants a GPU to be practical |
| ElevenLabs | Cloud | Yes | Cloud API / web; best-in-class voices and cloning |
| Amazon Polly | Cloud | Yes | Cloud API only |
| Google Cloud TTS | Cloud | Yes | Cloud API only |
| Azure AI Speech (TTS) | Cloud (+ container) | Yes* | Cloud API; disconnected container available |
| Murf AI | Cloud | Yes | Web studio; SaaS only |
| Play.ht | Cloud | Yes | Web / API; SaaS only |
| Resemble AI | Cloud (+ on-prem) | Yes* | Cloud API / web; on-prem on enterprise |
* = cloud by default; a self-hosted / on-prem option removes the internet dependency on paid enterprise tiers.
Platforms
Which operating systems / environments each tool runs on. Cloud APIs are marked “Any (HTTP)” because they are called over HTTPS from any platform; the row notes where a self-hostable component runs.
| Tool | Windows | macOS | Linux | Mobile / Embedded | Cloud |
|---|---|---|---|---|---|
| Piper | Yes | Yes | Yes | RPi / ARM | – |
| Coqui TTS | Yes | Yes | Yes | via export | – |
| eSpeak NG | Yes | Yes | Yes | Android | – |
| Kokoro | Yes | Yes | Yes | via ONNX | – |
| Bark | Yes | Yes | Yes | GPU recommended | – |
| ElevenLabs | Any (HTTP) | Any | Any | Any | Yes |
| Amazon Polly | Any (HTTP) | Any | Any | Any | Yes |
| Google Cloud TTS | Any (HTTP) | Any | Any | Any | Yes |
| Azure AI Speech (TTS) | Any (HTTP) | Any | Any | Any | Yes; container = Linux/Docker |
| Murf AI | Browser | Browser | Browser | – | Yes |
| Play.ht | Any (HTTP) | Any | Any | Any | Yes |
| Resemble AI | Any (HTTP) | Any | Any | Any | Yes; on-prem = Linux/Docker |
Licensing
| Tool | License | OSS | Cost Model |
|---|---|---|---|
| Piper | MIT | Yes | Free (self-host) |
| Coqui TTS | MPL-2.0 | Yes | Free (self-host); check per-voice model terms |
| eSpeak NG | GPL-3.0-or-later | Yes | Free (self-host) |
| Kokoro | Apache-2.0 | Yes | Free (self-host) |
| Bark | MIT | Yes | Free (self-host) |
| ElevenLabs | Proprietary | No | Per-character; freemium |
| Amazon Polly | Proprietary | No | Per-character usage |
| Google Cloud TTS | Proprietary | No | Per-character usage |
| Azure AI Speech (TTS) | Proprietary | No | Per-character; container tier |
| Murf AI | Proprietary | No | Per-seat subscription |
| Play.ht | Proprietary | No | Subscription; per-character API |
| Resemble AI | Proprietary | No | Usage; on-prem enterprise |
Feature Matrix
| Feature | Piper | Coqui TTS | eSpeak NG | Kokoro | Bark |
|---|---|---|---|---|---|
| Neural voice | Yes | Yes | No | Yes | Yes |
| Voice cloning | No | Yes (XTTS) | No | No | Partial |
| SSML / control | Basic | Partial | Yes | Basic | Prompt |
| Streaming | Yes | Partial | Yes | Partial | No |
| Languages | 30+ | 20+ | 100+ | few | 10+ |
| GPU needed | No | Optional | No | Optional | Yes |
| Feature | ElevenLabs | Polly | Azure | Murf | Play.ht | Resemble | |
|---|---|---|---|---|---|---|---|
| Neural voice | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Voice cloning | Yes | No | No | Yes | Partial | Yes | Yes |
| SSML / control | Partial | Yes | Yes | Yes | Studio | Partial | Yes |
| Streaming | Yes | Yes | Yes | Yes | No | Yes | Yes |
| Languages | 30+ | 30+ | 50+ | 100+ | 20+ | 30+ | 60+ |
| Self-host option | No | No | No | Yes | No | No | Yes |
SSO / OIDC Comparison
Local engines are libraries/CLIs with no user login; cloud APIs authenticate with API keys, and end-user SSO (where present) is an enterprise feature. The Authentik column notes fit with Authentik as the identity provider.
| Tool | End-user SSO | Authentik Notes |
|---|---|---|
| Piper / Coqui / eSpeak NG / Kokoro / Bark | N/A | Library / CLI; no login |
| ElevenLabs / Play.ht / Resemble AI | Paid | SSO on enterprise; API-key otherwise |
| Amazon Polly / Google / Azure | None | Cloud IAM, not Authentik OIDC |
| Murf AI | Paid | SSO on enterprise / team tiers |
Suitability Summary
For MSP Customers
- Privacy-first / air-gapped: self-hosted Piper (or eSpeak NG for footprint).
- Quality-first, cloud OK: ElevenLabs for the most natural voices and cloning; Polly / Google / Azure where the customer already lives in that cloud.
- Regulated but wants cloud quality: Resemble AI or Azure on-prem / container.
- Content studio for non-technical users: Murf AI or Play.ht.
Tools
12 tools.
Amazon Polly
Amazon Polly is AWS’s cloud text-to-speech service offering neural and standard voices across 30+ languages with SSML control, billed per character.
License: Proprietary (proprietary) · Kind: service · Deploy: saas · SSO: none · Tags: text-to-speech, tts, cloud, aws
Azure AI Speech (Text-to-Speech)
Azure AI Speech provides cloud text-to-speech with 400+ neural voices across 100+ languages, custom-voice cloning, and SSML; a disconnected container is available.
License: Proprietary (proprietary) · Kind: service · Deploy: saas, docker · SSO: none · Tags: text-to-speech, tts, cloud, azure, container
Bark
Bark is an open-source generative text-to-audio model from Suno that produces expressive speech and nonverbal sounds locally, best run with a GPU.
License: MIT (OSS) · Kind: library · Deploy: native · SSO: none · Tags: text-to-speech, tts, generative, offline
Coqui TTS
Coqui TTS is an open-source deep-learning text-to-speech toolkit with many pretrained models and XTTS voice cloning, running locally on CPU or GPU.
License: MPL-2.0 (OSS) · Kind: library · Deploy: native · SSO: none · Tags: text-to-speech, tts, offline, voice-cloning
ElevenLabs
ElevenLabs is a cloud text-to-speech and voice-AI platform known for highly natural voices, expressive control, and voice cloning across 30+ languages, via API and web.
License: Proprietary (proprietary) · Kind: service · Deploy: saas · SSO: OIDC, SAML · Tags: text-to-speech, tts, cloud, voice-cloning
eSpeak NG
eSpeak NG is a compact, fully offline formant-synthesis text-to-speech engine supporting 100+ languages; robotic but tiny, scriptable, and ideal for embedded use.
License: GPL-3.0-or-later (OSS) · Kind: cli · Deploy: native · SSO: none · Tags: text-to-speech, tts, offline, embedded
Google Cloud Text-to-Speech
Google Cloud Text-to-Speech is a cloud API synthesizing natural WaveNet and neural voices across 50+ languages with SSML control, billed per character.
License: Proprietary (proprietary) · Kind: service · Deploy: saas · SSO: none · Tags: text-to-speech, tts, cloud, gcp
Kokoro
Kokoro is a small 82M-parameter open-weight neural text-to-speech model delivering high-quality voices locally via ONNX or PyTorch, runnable on CPU.
License: Apache-2.0 (OSS) · Kind: library · Deploy: native · SSO: none · Tags: text-to-speech, tts, offline, neural
Murf AI
Murf AI is a cloud text-to-speech studio for creating voiceovers from a web app, with 120+ voices across 20+ languages, plus editing and collaboration features.
License: Proprietary (proprietary) · Kind: web · Deploy: saas · SSO: OIDC, SAML · Tags: text-to-speech, tts, cloud, voiceover
Piper
Piper is a fast, local neural text-to-speech engine optimized to run offline on modest hardware including Raspberry Pi, with many voices across 30+ languages.
License: MIT (OSS) · Kind: cli · Deploy: native · SSO: none · Tags: text-to-speech, tts, offline, neural
Play.ht
Play.ht is a cloud text-to-speech platform offering natural AI voices and voice cloning through a web studio and a streaming API, across 30+ languages.
License: Proprietary (proprietary) · Kind: web · Deploy: saas · SSO: OIDC, SAML · Tags: text-to-speech, tts, cloud, voice-cloning
Resemble AI
Resemble AI is a cloud voice-AI platform for text-to-speech and real-time voice cloning via API, with an on-premises deployment option for enterprises.
License: Proprietary (proprietary) · Kind: service · Deploy: saas, docker · SSO: OIDC, SAML · Tags: text-to-speech, tts, cloud, voice-cloning
