ListenSpeedScoresCapabilitiesHosted🗳 Vote ↗

Listen

One clip per model per voice mode. Audio is rig-independent (same weights → same output), so each sample is sourced once from the highest-fidelity available rig — Windows RTX 5090 where possible, else Linux, else Mac. The small tag on each row shows the source rig·device. Default voice = the model's own preset/built-in speaker; Cloning = the model imitating one reference voice (the house reference). Switch By prompt (compare every model on one sentence) and By model (audition one model across prompts) below; only one clip plays at a time. Speed per rig is on the Speed page.
Prompt 1 [en] "Open the browser and read my email."
Default voice (56)
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
AuroraTTS-100K95.6Kwin·cuda
Chatterbox1.2Bwin·cuda
Chatterbox Turbo744Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
Dia 1.6B-06261.6Bwin·cuda
dots.tts (soar) ⟳ voice varies2Blinux·cuda
DramaBox3.3Bwin·cuda
Higgs Audio v3 TTS ⟳ voice varies4Blinux·cuda
Hojo-TTS-Light 40M84.5M (40.4M LM + 13.3M fine + 30.8M decoder)win·cuda
Inflect-Micro v29.36Mwin·cuda
Inflect-Nano v23.96Mwin·cuda
KittenTTS 21.7Bwin·cuda
KittenTTS 2 (C++ CPU)1.7Bwin·cpu
KittenTTS Nano 0.1<100Mwin·cpu
KittenTTS Nano 0.815Mwin·cpu
Kokoro82Mwin·cuda
LFM2.5-Audio 1.5B1.5Bwin·cuda
LFM3-Audio 2B2.0Bwin·cuda
LinnetTTS41.4M (31.5M LM + 9.9M decoder)win·cpu
LongCat-AudioDiT 1B ⟳ voice varies1.42Bwin·cuda
LongCat-AudioDiT 3.5B ⟳ voice varies3.83Bwin·cuda
Magpie-TTS357Mwin·cuda
Mars5-TTS1.2Bwin·cuda
Maya13Bwin·cuda
MeloTTS~52Mwin·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
MiraTTS0.5Blinux·cuda
Miso TTS 8B ⚠ artifacts8.2Bwin·cuda
OmniVoice~1Bwin·cuda
Orpheus-TTS 3B3.3Blinux·cuda
OuteTTS 1.0 1B1Bwin·cuda
Parler-TTS Mini v1878Mwin·cuda
Phonon 31m41.1M (30.8M LM + Mimi)win·cpu
Phonon 40m50.5M (40.2M LM + Mimi)win·cpu
Piper~25MBwin·cpu
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 0.6B CustomVoice0.6Bwin·cuda
sanoTTS Amy1.46Mwin·cpu
sanoTTS Heart-Nano294Kwin·cpu
Scylla's Band~131Mwin·cpu
Sesame CSM-1B1Bwin·cuda
Soprano 1.1 80M80Mwin·cuda
SteerTTS40.8Mwin·cuda
SteerTTS (no refiner)23.1Mwin·cuda
StyleTTS 2~148Mwin·cuda
Supertonic 399Mwin·cpu
SupraTTS 0.1 Beta43.5Mwin·cuda
Vaniq-Edge ⚠ onset words8.91Mwin·cuda
VibeVoice Realtime 0.5B0.5Bwin·cuda
VoxCPM2 2B2Bwin·cuda
Voxtral 4B TTS4Bmac·mps
Cloning — House reference (60)
▶ Reference voice — the target each clone imitates:
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
AuK1.53Bwin·cuda
AuK-Flash1.53Bwin·cuda
Breeze TTS 23.47Bwin·cuda
Chatterbox1.2Bwin·cuda
Chatterbox Turbo744Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
CosyVoice 3 0.5B ⟳ length varies0.5Blinux·cuda
Dia 1.6B-06261.6Bwin·cuda
dots.tts (soar)2Blinux·cuda
DramaBox3.3Bwin·cuda
Echo-TTS2.8Bwin·cuda
F5-TTS v1330Mwin·cuda
FireRedTTS32.12Bwin·cuda
Fish Audio Drama 3 (ComfyUI port)35.9B (MoE, 3B active)win·cuda
Fish Speech 1.5~500Mwin·cuda
Fish Speech S2-Pro4Blinux·cuda
Higgs Audio v3 TTS4Blinux·cuda
Hojo-TTS-Light 80M ⚠ clone doesn't hold ref167M (89.4M LM + 20.3M codec enc + 45.0M dec + 12.0M spk)win·cuda
IndexTTS-21.5Bwin·cuda
IndexTTS-2.50.97Bwin·cuda
KittenTTS 21.7Bwin·cuda
LongCat-AudioDiT 1B1.42Bwin·cuda
LongCat-AudioDiT 3.5B ⟳ clone less reliable3.83Bwin·cuda
LuxTTS123Mlinux·cuda
Mars5-TTS1.2Bwin·cuda
MetaVoice-1B1.2Blinux·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
MioTTS 0.1B0.1Blinux·cuda
MioTTS 0.6B ⟳ length varies0.6Blinux·cuda
MiraTTS0.5Bwin·cuda
Miso TTS 8B ⟳ clone unstable8.2Bwin·cuda
MOSS-TTS v1.08Bwin·cuda
MOSS-TTS v1.58Bwin·cuda
MOSS-TTS-Nano100Mwin·cuda
NeuTTS Air748Mwin·cuda
NeuTTS Nano229Mwin·cuda
OmniVoice~1Bwin·cuda
OpenVoice v2~100Mwin·cuda
OuteTTS 1.0 1B1Bwin·cuda
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 1.7B (CUDA-graph)1.7Bwin·cuda
Sesame CSM-1B1Bwin·cuda
Sopro V2 Turbo120Mwin·cuda
Sopro V2 Turbo (streaming)120Mwin·cuda
SteerTTS40.8Mwin·cuda
SteerTTS (no refiner)23.1Mwin·cuda
Step-Audio-EditX3Blinux·cuda
StyleTTS 2~148Mwin·cuda
VibeVoice 1.5B1.5Bwin·cuda
VibeVoice 7B7Bwin·cuda
VoxCPM2 2B2Bwin·cuda
WavTTS 0.67B0.67Bwin·cuda
ZipVoice 123M123Mwin·cuda
Zonos v0.11.6Bwin·cuda
Zonos28B (MoE, ~900M active)linux·cuda
Prompt 2 [en] "I'll start a new git branch, push the changes, and open a pull request when the tests pass."
Default voice (56)
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
AuroraTTS-100K95.6Kwin·cuda
Chatterbox1.2Bwin·cuda
Chatterbox Turbo744Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
Dia 1.6B-06261.6Bwin·cuda
dots.tts (soar) ⟳ voice varies2Blinux·cuda
DramaBox3.3Bwin·cuda
Higgs Audio v3 TTS ⟳ voice varies4Blinux·cuda
Hojo-TTS-Light 40M84.5M (40.4M LM + 13.3M fine + 30.8M decoder)win·cuda
Inflect-Micro v29.36Mwin·cuda
Inflect-Nano v23.96Mwin·cuda
KittenTTS 21.7Bwin·cuda
KittenTTS 2 (C++ CPU)1.7Bwin·cpu
KittenTTS Nano 0.1<100Mwin·cpu
KittenTTS Nano 0.815Mwin·cpu
Kokoro82Mwin·cuda
LFM2.5-Audio 1.5B1.5Bwin·cuda
LFM3-Audio 2B2.0Bwin·cuda
LinnetTTS41.4M (31.5M LM + 9.9M decoder)win·cpu
LongCat-AudioDiT 1B ⟳ voice varies1.42Bwin·cuda
LongCat-AudioDiT 3.5B ⟳ voice varies3.83Bwin·cuda
Magpie-TTS357Mwin·cuda
Mars5-TTS1.2Bwin·cuda
Maya13Bwin·cuda
MeloTTS~52Mwin·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
MiraTTS0.5Blinux·cuda
Miso TTS 8B ⚠ artifacts8.2Bwin·cuda
OmniVoice~1Bwin·cuda
Orpheus-TTS 3B3.3Blinux·cuda
OuteTTS 1.0 1B1Bwin·cuda
Parler-TTS Mini v1878Mwin·cuda
Phonon 31m41.1M (30.8M LM + Mimi)win·cpu
Phonon 40m50.5M (40.2M LM + Mimi)win·cpu
Piper~25MBwin·cpu
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 0.6B CustomVoice0.6Bwin·cuda
sanoTTS Amy1.46Mwin·cpu
sanoTTS Heart-Nano294Kwin·cpu
Scylla's Band~131Mwin·cpu
Sesame CSM-1B1Bwin·cuda
Soprano 1.1 80M80Mwin·cuda
SteerTTS40.8Mwin·cuda
SteerTTS (no refiner)23.1Mwin·cuda
StyleTTS 2~148Mwin·cuda
Supertonic 399Mwin·cpu
SupraTTS 0.1 Beta43.5Mwin·cuda
Vaniq-Edge ⚠ onset words8.91Mwin·cuda
VibeVoice Realtime 0.5B0.5Bwin·cuda
VoxCPM2 2B2Bwin·cuda
Voxtral 4B TTS4Bmac·mps
Cloning — House reference (60)
▶ Reference voice — the target each clone imitates:
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
AuK1.53Bwin·cuda
AuK-Flash1.53Bwin·cuda
Breeze TTS 23.47Bwin·cuda
Chatterbox1.2Bwin·cuda
Chatterbox Turbo744Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
CosyVoice 3 0.5B ⟳ length varies0.5Blinux·cuda
Dia 1.6B-06261.6Bwin·cuda
dots.tts (soar)2Blinux·cuda
DramaBox3.3Bwin·cuda
Echo-TTS2.8Bwin·cuda
F5-TTS v1330Mwin·cuda
FireRedTTS32.12Bwin·cuda
Fish Audio Drama 3 (ComfyUI port)35.9B (MoE, 3B active)win·cuda
Fish Speech 1.5~500Mwin·cuda
Fish Speech S2-Pro4Blinux·cuda
Higgs Audio v3 TTS4Blinux·cuda
Hojo-TTS-Light 80M ⚠ clone doesn't hold ref167M (89.4M LM + 20.3M codec enc + 45.0M dec + 12.0M spk)win·cuda
IndexTTS-21.5Bwin·cuda
IndexTTS-2.50.97Bwin·cuda
KittenTTS 21.7Bwin·cuda
LongCat-AudioDiT 1B1.42Bwin·cuda
LongCat-AudioDiT 3.5B ⟳ clone less reliable3.83Bwin·cuda
LuxTTS123Mlinux·cuda
Mars5-TTS1.2Bwin·cuda
MetaVoice-1B1.2Blinux·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
MioTTS 0.1B0.1Blinux·cuda
MioTTS 0.6B ⟳ length varies0.6Blinux·cuda
MiraTTS0.5Bwin·cuda
Miso TTS 8B ⟳ clone unstable8.2Bwin·cuda
MOSS-TTS v1.08Bwin·cuda
MOSS-TTS v1.58Bwin·cuda
MOSS-TTS-Nano100Mwin·cuda
NeuTTS Air748Mwin·cuda
NeuTTS Nano229Mwin·cuda
OmniVoice~1Bwin·cuda
OpenVoice v2~100Mwin·cuda
OuteTTS 1.0 1B1Bwin·cuda
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 1.7B (CUDA-graph)1.7Bwin·cuda
Sesame CSM-1B1Bwin·cuda
Sopro V2 Turbo120Mwin·cuda
Sopro V2 Turbo (streaming)120Mwin·cuda
SteerTTS40.8Mwin·cuda
SteerTTS (no refiner)23.1Mwin·cuda
Step-Audio-EditX3Blinux·cuda
StyleTTS 2~148Mwin·cuda
VibeVoice 1.5B1.5Bwin·cuda
VibeVoice 7B7Bwin·cuda
VoxCPM2 2B2Bwin·cuda
WavTTS 0.67B0.67Bwin·cuda
ZipVoice 123M123Mwin·cuda
Zonos v0.11.6Bwin·cuda
Zonos28B (MoE, ~900M active)linux·cuda
Prompt 3 [en] "The Parakeet TDT zero point six billion parameter model achieves one point six nine percent word error rate on LibriSpeech test-clean, beating Whisper Large V3 at two point seven percent while running at over two thousand times realtime on a single GPU."
Default voice (55)
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
AuroraTTS-100K95.6Kwin·cuda
Chatterbox1.2Bwin·cuda
Chatterbox Turbo744Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
Dia 1.6B-06261.6Bwin·cuda
dots.tts (soar) ⟳ voice varies2Blinux·cuda
DramaBox3.3Bwin·cuda
Higgs Audio v3 TTS ⟳ voice varies4Blinux·cuda
Hojo-TTS-Light 40M84.5M (40.4M LM + 13.3M fine + 30.8M decoder)win·cuda
Inflect-Micro v29.36Mwin·cuda
Inflect-Nano v23.96Mwin·cuda
KittenTTS 21.7Bwin·cuda
KittenTTS 2 (C++ CPU)1.7Bwin·cpu
KittenTTS Nano 0.1<100Mwin·cpu
KittenTTS Nano 0.815Mwin·cpu
Kokoro82Mwin·cuda
LFM2.5-Audio 1.5B1.5Bwin·cuda
LFM3-Audio 2B2.0Bwin·cuda
LinnetTTS41.4M (31.5M LM + 9.9M decoder)win·cpu
LongCat-AudioDiT 1B ⟳ voice varies1.42Bwin·cuda
LongCat-AudioDiT 3.5B ⟳ voice varies3.83Bwin·cuda
Magpie-TTS357Mwin·cuda
Mars5-TTS1.2Bwin·cuda
Maya13Bwin·cuda
MeloTTS~52Mwin·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
MiraTTS0.5Blinux·cuda
Miso TTS 8B ⚠ artifacts8.2Bwin·cuda
OmniVoice~1Bwin·cuda
Orpheus-TTS 3B3.3Blinux·cuda
OuteTTS 1.0 1B1Bwin·cuda
Parler-TTS Mini v1878Mwin·cuda
Phonon 31m41.1M (30.8M LM + Mimi)win·cpu
Phonon 40m50.5M (40.2M LM + Mimi)win·cpu
Piper~25MBwin·cpu
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 0.6B CustomVoice0.6Bwin·cuda
sanoTTS Amy1.46Mwin·cpu
Scylla's Band~131Mwin·cpu
Sesame CSM-1B1Bwin·cuda
Soprano 1.1 80M80Mwin·cuda
SteerTTS40.8Mwin·cuda
SteerTTS (no refiner)23.1Mwin·cuda
StyleTTS 2~148Mwin·cuda
Supertonic 399Mwin·cpu
SupraTTS 0.1 Beta43.5Mwin·cuda
Vaniq-Edge ⚠ onset words8.91Mwin·cuda
VibeVoice Realtime 0.5B0.5Bwin·cuda
VoxCPM2 2B2Bwin·cuda
Voxtral 4B TTS4Bmac·mps
Cloning — House reference (60)
▶ Reference voice — the target each clone imitates:
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
AuK1.53Bwin·cuda
AuK-Flash1.53Bwin·cuda
Breeze TTS 23.47Bwin·cuda
Chatterbox1.2Bwin·cuda
Chatterbox Turbo744Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
CosyVoice 3 0.5B ⟳ length varies0.5Blinux·cuda
Dia 1.6B-06261.6Bwin·cuda
dots.tts (soar)2Blinux·cuda
DramaBox3.3Bwin·cuda
Echo-TTS2.8Bwin·cuda
F5-TTS v1330Mwin·cuda
FireRedTTS32.12Bwin·cuda
Fish Audio Drama 3 (ComfyUI port)35.9B (MoE, 3B active)win·cuda
Fish Speech 1.5~500Mwin·cuda
Fish Speech S2-Pro4Blinux·cuda
Higgs Audio v3 TTS4Blinux·cuda
Hojo-TTS-Light 80M ⚠ clone doesn't hold ref167M (89.4M LM + 20.3M codec enc + 45.0M dec + 12.0M spk)win·cuda
IndexTTS-21.5Bwin·cuda
IndexTTS-2.50.97Bwin·cuda
KittenTTS 21.7Bwin·cuda
LongCat-AudioDiT 1B1.42Bwin·cuda
LongCat-AudioDiT 3.5B ⟳ clone less reliable3.83Bwin·cuda
LuxTTS123Mlinux·cuda
Mars5-TTS1.2Bwin·cuda
MetaVoice-1B1.2Blinux·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
MioTTS 0.1B0.1Blinux·cuda
MioTTS 0.6B ⟳ length varies0.6Blinux·cuda
MiraTTS0.5Bwin·cuda
Miso TTS 8B ⟳ clone unstable8.2Bwin·cuda
MOSS-TTS v1.08Bwin·cuda
MOSS-TTS v1.58Bwin·cuda
MOSS-TTS-Nano100Mwin·cuda
NeuTTS Air748Mwin·cuda
NeuTTS Nano229Mwin·cuda
OmniVoice~1Bwin·cuda
OpenVoice v2~100Mwin·cuda
OuteTTS 1.0 1B1Bwin·cuda
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 1.7B (CUDA-graph)1.7Bwin·cuda
Sesame CSM-1B1Bwin·cuda
Sopro V2 Turbo120Mwin·cuda
Sopro V2 Turbo (streaming)120Mwin·cuda
SteerTTS40.8Mwin·cuda
SteerTTS (no refiner)23.1Mwin·cuda
Step-Audio-EditX3Blinux·cuda
StyleTTS 2~148Mwin·cuda
VibeVoice 1.5B1.5Bwin·cuda
VibeVoice 7B7Bwin·cuda
VoxCPM2 2B2Bwin·cuda
WavTTS 0.67B0.67Bwin·cuda
ZipVoice 123M123Mwin·cuda
Zonos v0.11.6Bwin·cuda
Zonos28B (MoE, ~900M active)linux·cuda
Prompt 4 [en] "Run pytest tests slash test underscore voice dot py with verbose flag and capture flag set to no."
Default voice (56)
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
AuroraTTS-100K95.6Kwin·cuda
Chatterbox1.2Bwin·cuda
Chatterbox Turbo744Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
Dia 1.6B-06261.6Bwin·cuda
dots.tts (soar) ⟳ voice varies2Blinux·cuda
DramaBox3.3Bwin·cuda
Higgs Audio v3 TTS ⟳ voice varies4Blinux·cuda
Hojo-TTS-Light 40M84.5M (40.4M LM + 13.3M fine + 30.8M decoder)win·cuda
Inflect-Micro v29.36Mwin·cuda
Inflect-Nano v23.96Mwin·cuda
KittenTTS 21.7Bwin·cuda
KittenTTS 2 (C++ CPU)1.7Bwin·cpu
KittenTTS Nano 0.1<100Mwin·cpu
KittenTTS Nano 0.815Mwin·cpu
Kokoro82Mwin·cuda
LFM2.5-Audio 1.5B1.5Bwin·cuda
LFM3-Audio 2B2.0Bwin·cuda
LinnetTTS41.4M (31.5M LM + 9.9M decoder)win·cpu
LongCat-AudioDiT 1B ⟳ voice varies1.42Bwin·cuda
LongCat-AudioDiT 3.5B ⟳ voice varies3.83Bwin·cuda
Magpie-TTS357Mwin·cuda
Mars5-TTS1.2Bwin·cuda
Maya13Bwin·cuda
MeloTTS~52Mwin·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
MiraTTS0.5Blinux·cuda
Miso TTS 8B ⚠ artifacts8.2Bwin·cuda
OmniVoice~1Bwin·cuda
Orpheus-TTS 3B3.3Blinux·cuda
OuteTTS 1.0 1B1Bwin·cuda
Parler-TTS Mini v1878Mwin·cuda
Phonon 31m41.1M (30.8M LM + Mimi)win·cpu
Phonon 40m50.5M (40.2M LM + Mimi)win·cpu
Piper~25MBwin·cpu
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 0.6B CustomVoice0.6Bwin·cuda
sanoTTS Amy1.46Mwin·cpu
sanoTTS Heart-Nano294Kwin·cpu
Scylla's Band~131Mwin·cpu
Sesame CSM-1B1Bwin·cuda
Soprano 1.1 80M80Mwin·cuda
SteerTTS40.8Mwin·cuda
SteerTTS (no refiner)23.1Mwin·cuda
StyleTTS 2~148Mwin·cuda
Supertonic 399Mwin·cpu
SupraTTS 0.1 Beta43.5Mwin·cuda
Vaniq-Edge ⚠ onset words8.91Mwin·cuda
VibeVoice Realtime 0.5B0.5Bwin·cuda
VoxCPM2 2B2Bwin·cuda
Voxtral 4B TTS4Bmac·mps
Cloning — House reference (60)
▶ Reference voice — the target each clone imitates:
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
AuK1.53Bwin·cuda
AuK-Flash1.53Bwin·cuda
Breeze TTS 23.47Bwin·cuda
Chatterbox1.2Bwin·cuda
Chatterbox Turbo744Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
CosyVoice 3 0.5B ⟳ length varies0.5Blinux·cuda
Dia 1.6B-06261.6Bwin·cuda
dots.tts (soar)2Blinux·cuda
DramaBox3.3Bwin·cuda
Echo-TTS2.8Bwin·cuda
F5-TTS v1330Mwin·cuda
FireRedTTS32.12Bwin·cuda
Fish Audio Drama 3 (ComfyUI port)35.9B (MoE, 3B active)win·cuda
Fish Speech 1.5~500Mwin·cuda
Fish Speech S2-Pro4Blinux·cuda
Higgs Audio v3 TTS4Blinux·cuda
Hojo-TTS-Light 80M ⚠ clone doesn't hold ref167M (89.4M LM + 20.3M codec enc + 45.0M dec + 12.0M spk)win·cuda
IndexTTS-21.5Bwin·cuda
IndexTTS-2.50.97Bwin·cuda
KittenTTS 21.7Bwin·cuda
LongCat-AudioDiT 1B1.42Bwin·cuda
LongCat-AudioDiT 3.5B ⟳ clone less reliable3.83Bwin·cuda
LuxTTS123Mlinux·cuda
Mars5-TTS1.2Bwin·cuda
MetaVoice-1B1.2Blinux·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
MioTTS 0.1B0.1Blinux·cuda
MioTTS 0.6B ⟳ length varies0.6Blinux·cuda
MiraTTS0.5Bwin·cuda
Miso TTS 8B ⟳ clone unstable8.2Bwin·cuda
MOSS-TTS v1.08Bwin·cuda
MOSS-TTS v1.58Bwin·cuda
MOSS-TTS-Nano100Mwin·cuda
NeuTTS Air748Mwin·cuda
NeuTTS Nano229Mwin·cuda
OmniVoice~1Bwin·cuda
OpenVoice v2~100Mwin·cuda
OuteTTS 1.0 1B1Bwin·cuda
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 1.7B (CUDA-graph)1.7Bwin·cuda
Sesame CSM-1B1Bwin·cuda
Sopro V2 Turbo120Mwin·cuda
Sopro V2 Turbo (streaming)120Mwin·cuda
SteerTTS40.8Mwin·cuda
SteerTTS (no refiner)23.1Mwin·cuda
Step-Audio-EditX3Blinux·cuda
StyleTTS 2~148Mwin·cuda
VibeVoice 1.5B1.5Bwin·cuda
VibeVoice 7B7Bwin·cuda
VoxCPM2 2B2Bwin·cuda
WavTTS 0.67B0.67Bwin·cuda
ZipVoice 123M123Mwin·cuda
Zonos v0.11.6Bwin·cuda
Zonos28B (MoE, ~900M active)linux·cuda
Prompt 5 [fr] "Bonjour, je m'appelle Cicero et je vais vous aider avec votre code aujourd'hui."
Default voice (25)
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
dots.tts (soar) ⟳ voice varies2Blinux·cuda
Higgs Audio v3 TTS ⟳ voice varies4Blinux·cuda
KittenTTS 21.7Bwin·cuda
KittenTTS 2 (C++ CPU)1.7Bwin·cpu
Kokoro82Mwin·cuda
LFM3-Audio 2B2.0Bwin·cuda
Magpie-TTS357Mwin·cuda
MeloTTS~52Mwin·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
OmniVoice~1Bwin·cuda
OuteTTS 1.0 1B1Bwin·cuda
Phonon 40m50.5M (40.2M LM + Mimi)win·cpu
Piper~25MBwin·cpu
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 0.6B CustomVoice0.6Bwin·cuda
Scylla's Band~131Mwin·cpu
Supertonic 399Mwin·cpu
VoxCPM2 2B2Bwin·cuda
Voxtral 4B TTS4Bmac·mps
Cloning — House reference (30)
▶ Reference voice — the target each clone imitates:
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
AuK1.53Bwin·cuda
AuK-Flash1.53Bwin·cuda
Coqui XTTS-v2750Mwin·cuda
CosyVoice 3 0.5B ⟳ length varies0.5Blinux·cuda
dots.tts (soar)2Blinux·cuda
Fish Audio Drama 3 (ComfyUI port)35.9B (MoE, 3B active)win·cuda
Fish Speech 1.5~500Mwin·cuda
Fish Speech S2-Pro4Blinux·cuda
Higgs Audio v3 TTS4Blinux·cuda
KittenTTS 21.7Bwin·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Ming-3.0-TTS tiny8.98B (MoE, 1.3B active)linux·cuda
MOSS-TTS v1.08Bwin·cuda
MOSS-TTS v1.58Bwin·cuda
MOSS-TTS-Nano100Mwin·cuda
NeuTTS Nano229Mwin·cuda
OmniVoice~1Bwin·cuda
OpenVoice v2~100Mwin·cuda
OuteTTS 1.0 1B1Bwin·cuda
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 1.7B (CUDA-graph)1.7Bwin·cuda
Sopro V2 Turbo120Mwin·cuda
Sopro V2 Turbo (streaming)120Mwin·cuda
VoxCPM2 2B2Bwin·cuda
ZipVoice 123M123Mwin·cuda
Zonos v0.11.6Bwin·cuda
Prompt 6 [es] "¿Este modelo es bilingüe? Añade el español, corrige la pronunciación automática y reinicia el servidor."
Default voice (16)
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
KittenTTS 21.7Bwin·cuda
KittenTTS 2 (C++ CPU)1.7Bwin·cpu
Kokoro82Mwin·cuda
LFM3-Audio 2B2.0Blinux·cuda
MeloTTS~52Mwin·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
Phonon 40m50.5M (40.2M LM + Mimi)win·cpu
Piper~25MBwin·cpu
Pocket-TTS v4100Mwin·cpu
Scylla's Band~131Mlinux·cpu
Supertonic 399Mwin·cpu
Cloning — House reference (13)
▶ Reference voice — the target each clone imitates:
ModelSizeSourceAudio
Audio8 TTS 0.6B (compiled)601Mwin·cuda
Audio8 TTS Preview 0.1B170Mwin·cuda
Audio8 TTS Preview 0.6B601Mwin·cuda
Audio8 TTS Realistic 0.6B601Mwin·cuda
Coqui XTTS-v2750Mwin·cuda
Fish Audio Drama 3 (ComfyUI port)35.9B (MoE, 3B active)win·cuda
IndexTTS-2.50.97Bwin·cuda
KittenTTS 21.7Bwin·cuda
Ming-3.0-TTS 0.5B1.48Bwin·cuda
NeuTTS Nano229Mwin·cuda
OpenVoice v2~100Mwin·cuda
Pocket-TTS v4100Mwin·cpu
Qwen3-TTS 1.7B (CUDA-graph)1.7Bwin·cuda