Voice catalog

Voice catalog

Flat reference of every voice the app ships in VoiceCatalog.kt as of v1.2.4. 136 voices total across five engines (four local + one cloud):

• Piper (in-process, local) — 45 voices, single-speaker, ~14–30 MB each.
• Kokoro (in-process, local) — 53 voices in one ~330 MB multi-speaker model.
• KittenTTS (in-process, local — shipped v0.5.36, #119) — 8 en_US speakers in one ~24 MB shared model. Lightest tier.
• Azure HD (remote, BYOK) — 20 voices, no local download, billed by Microsoft to your key.
• Supertonic 3 (in-process, local — newest, live since v1.2.3) — 10 en_US speakers (5F / 5M), High quality, from a ~139 MB shared model downloaded on first use (#1191 / #1237).

The narrative version of this catalog lives at the website's Voices page and the wiki's Voices page. This page is the lookup table.

How IDs are structured

Voice IDs follow <engine>_<name>_<lang>_<tier> (Piper) or <engine>_<name>_<lang>_<speakerId> (Kokoro) or <engine>_<name>_<lang>_<tierMarker> (Azure). The _int8 suffix on some Piper voices is historical — kept stable so existing installs don't have to re-pick a voice. Files served from voices-v2 use fp32 weights regardless of ID suffix.

Featured voices (highlighted with ⭐ in the picker, shown first to new users): Lessac (en_US Piper, high), Cori (en_GB Piper, high), Aoede (Kokoro, en_US, F, speaker 1).

Piper (45 voices)

Single-speaker neural voices from the rhasspy/piper project. Each voice ships as {lang}-{voice}-{tier}.onnx + .tokens.txt from the voices-v2 GitHub release. Quality tiers: low (smallest, fastest), medium (default), high (most natural, slowest).

en_US (29 voices)

ID	Name	Tier	Size
piper_lessac_en_US_high	⭐ Lessac	High	113.9 MB
piper_lessac_en_US_medium	Lessac	Medium	63.1 MB
piper_lessac_en_US_low	Lessac	Low	63.1 MB
piper_ryan_en_US_high	⭐ Ryan	High	120.8 MB
piper_ryan_en_US_medium	Ryan	Medium	63.1 MB
piper_ryan_en_US_low	Ryan	Low	63.1 MB
piper_amy_en_US_medium	⭐ Amy	Medium	63.2 MB
piper_amy_en_US_low	Amy	Low	63.1 MB
piper_arctic_en_US_medium	Arctic	Medium	76.7 MB
piper_bryce_en_US_medium	Bryce	Medium	63.2 MB
piper_danny_en_US_low	Danny	Low	63.1 MB
piper_glados_en_US_high	GLaDOS	High	113.8 MB
piper_glados_en_US_medium	GLaDOS	Medium	63.5 MB
piper_hfc_female_en_US_medium	HFC (female)	Medium	63.1 MB
piper_hfc_male_en_US_medium	HFC (male)	Medium	63.1 MB
piper_joe_en_US_medium	Joe	Medium	63.1 MB
piper_john_en_US_medium	John	Medium	63.2 MB
piper_kathleen_en_US_low	Kathleen	Low	63.1 MB
piper_kristin_en_US_medium	Kristin	Medium	63.5 MB
piper_kusal_en_US_medium	Kusal	Medium	63.2 MB
piper_l2arctic_en_US_medium	L2 Arctic	Medium	76.8 MB
piper_libritts_en_US_high	LibriTTS	High	129.4 MB
piper_libritts_r_en_US_medium	LibriTTS R	Medium	78.5 MB
piper_ljspeech_en_US_high	LJ Speech	High	114.2 MB
piper_ljspeech_en_US_medium	LJ Speech	Medium	63.5 MB
piper_miro_en_US_high	Miro	High	63.5 MB
piper_norman_en_US_medium	Norman	Medium	63.5 MB
piper_reza_ibrahim_en_US_medium	Reza Ibrahim	Medium	63.5 MB
piper_sam_en_US_medium	Sam	Medium	62.9 MB

en_GB (16 voices)

ID	Name	Tier	Size
piper_alan_en_GB_low	Alan	Low	63.1 MB
piper_alan_en_GB_medium	Alan	Medium	63.2 MB
piper_alba_en_GB_medium	Alba	Medium	63.2 MB
piper_aru_en_GB_medium	Aru	Medium	76.8 MB
piper_cori_en_GB_high	Cori (⭐ featured)	High	114.2 MB
piper_cori_en_GB_medium	Cori	Medium	63.5 MB
piper_dii_en_GB_high	Dii	High	63.5 MB
piper_jenny_dioco_en_GB_medium	Jenny Dioco	Medium	63.2 MB
piper_miro_en_GB_high	Miro	High	63.5 MB
piper_northern_english_male_en_GB_medium	Northern English (male)	Medium	63.2 MB
piper_semaine_en_GB_medium	Semaine	Medium	76.7 MB
piper_southern_english_female_en_GB_low	Southern English (female)	Low	63.1 MB
piper_southern_english_female_en_GB_medium	Southern English (female)	Medium	77.1 MB
piper_southern_english_male_en_GB_medium	Southern English (male)	Medium	77.1 MB
piper_sweetbbak_amy_en_GB_high	Sweetbbak Amy	High	114.2 MB
piper_vctk_en_GB_medium	VCTK	Medium	77.0 MB

Kokoro (53 voices)

Multi-speaker model from hexgrad/kokoro. One ~330 MB download (kokoro-multi-lang-v1_1) provides all 53 speakers. Pick by speaker ID inside the model.

English (28)

ID	Name	Lang	Gender	Speaker
kokoro_alloy_en_US_0	Alloy	en_US	F	0
kokoro_aoede_en_US_1	⭐ Aoede	en_US	F	1
kokoro_bella_en_US_2	Bella	en_US	F	2
kokoro_heart_en_US_3	Heart	en_US	F	3
kokoro_jessica_en_US_4	Jessica	en_US	F	4
kokoro_kore_en_US_5	Kore	en_US	F	5
kokoro_nicole_en_US_6	Nicole	en_US	F	6
kokoro_nova_en_US_7	Nova	en_US	F	7
kokoro_river_en_US_8	River	en_US	F	8
kokoro_sarah_en_US_9	Sarah	en_US	F	9
kokoro_sky_en_US_10	Sky	en_US	F	10
kokoro_adam_en_US_11	Adam	en_US	M	11
kokoro_echo_en_US_12	Echo	en_US	M	12
kokoro_eric_en_US_13	Eric	en_US	M	13
kokoro_fenrir_en_US_14	Fenrir	en_US	M	14
kokoro_liam_en_US_15	Liam	en_US	M	15
kokoro_michael_en_US_16	Michael	en_US	M	16
kokoro_onyx_en_US_17	Onyx	en_US	M	17
kokoro_puck_en_US_18	Puck	en_US	M	18
kokoro_santa_en_US_19	Santa	en_US	M	19
kokoro_alice_en_GB_20	Alice	en_GB	F	20
kokoro_emma_en_GB_21	Emma	en_GB	F	21
kokoro_isabella_en_GB_22	Isabella	en_GB	F	22
kokoro_lily_en_GB_23	Lily	en_GB	F	23
kokoro_daniel_en_GB_24	Daniel	en_GB	M	24
kokoro_fable_en_GB_25	Fable	en_GB	M	25
kokoro_george_en_GB_26	George	en_GB	M	26
kokoro_lewis_en_GB_27	Lewis	en_GB	M	27

Other languages (25)

ID	Name	Lang	Gender	Speaker
kokoro_dora_es_ES_28	Dora	es_ES	F	28
kokoro_alex_es_ES_29	Alex	es_ES	M	29
kokoro_siwis_fr_FR_30	Siwis	fr_FR	F	30
kokoro_alpha_hi_IN_31	Alpha	hi_IN	F	31
kokoro_beta_hi_IN_32	Beta	hi_IN	F	32
kokoro_omega_hi_IN_33	Omega	hi_IN	M	33
kokoro_psi_hi_IN_34	Psi	hi_IN	M	34
kokoro_sara_it_IT_35	Sara	it_IT	F	35
kokoro_nicola_it_IT_36	Nicola	it_IT	M	36
kokoro_alpha_ja_JP_37	Alpha	ja_JP	F	37
kokoro_gongitsune_ja_JP_38	Gongitsune	ja_JP	F	38
kokoro_nezumi_ja_JP_39	Nezumi	ja_JP	F	39
kokoro_tebukuro_ja_JP_40	Tebukuro	ja_JP	F	40
kokoro_kumo_ja_JP_41	Kumo	ja_JP	M	41
kokoro_dora_pt_PT_42	Dora	pt_PT	F	42
kokoro_alex_pt_PT_43	Alex	pt_PT	M	43
kokoro_santa_pt_PT_44	Santa	pt_PT	M	44
kokoro_xiaobei_zh_CN_45	Xiaobei	zh_CN	F	45
kokoro_xiaoni_zh_CN_46	Xiaoni	zh_CN	F	46
kokoro_xiaoxiao_zh_CN_47	Xiaoxiao	zh_CN	F	47
kokoro_xiaoyi_zh_CN_48	Xiaoyi	zh_CN	F	48
kokoro_yunjian_zh_CN_49	Yunjian	zh_CN	M	49
kokoro_yunxi_zh_CN_50	Yunxi	zh_CN	M	50
kokoro_yunxia_zh_CN_51	Yunxia	zh_CN	M	51
kokoro_yunyang_zh_CN_52	Yunyang	zh_CN	M	52

Azure HD (20 voices, BYOK)

Azure Cognitive Services HD voices. Bring your own key + region (default region: eastus). Candela does not bill you — Azure does, at the rate published in your subscription. Catalog rate-card stored in VoiceCost.centsPer1MChars is 3000 (~$30 / 1M characters); update there if Microsoft's pricing changes.

Dragon HD — Azure's 2025 generative tier (4)

ID	Name	Lang	Azure voice
azure_ava_en_US_dragon_hd	Ava (Dragon HD)	en_US	en-US-AvaDragonHDLatestNeural
azure_andrew_en_US_dragon_hd	Andrew (Dragon HD)	en_US	en-US-AndrewDragonHDLatestNeural
azure_brian_en_US_dragon_hd	Brian (Dragon HD)	en_US	en-US-BrianMultilingualNeural
azure_emma_en_US_dragon_hd	Emma (Dragon HD)	en_US	en-US-EmmaMultilingualNeural

en-US HD Neural (8)

ID	Name	Azure voice
azure_aria_en_US_hd	Aria (HD Neural)	en-US-AriaNeural
azure_jenny_en_US_hd	Jenny (HD Neural)	en-US-JennyMultilingualNeural
azure_guy_en_US_hd	Guy (HD Neural)	en-US-GuyNeural
azure_davis_en_US_hd	Davis (HD Neural)	en-US-DavisMultilingualNeural
azure_tony_en_US_hd	Tony (HD Neural)	en-US-TonyNeural
azure_sara_en_US_hd	Sara (HD Neural)	en-US-SaraNeural
azure_christopher_en_US_hd	Christopher (HD Neural)	en-US-ChristopherNeural
azure_nancy_en_US_hd	Nancy (HD Neural)	en-US-NancyNeural

Other English locales (8)

ID	Name	Lang	Azure voice
azure_sonia_en_GB_hd	Sonia (British)	en_GB	en-GB-SoniaNeural
azure_ryan_en_GB_hd	Ryan (British)	en_GB	en-GB-RyanNeural
azure_libby_en_GB_hd	Libby (British)	en_GB	en-GB-LibbyNeural
azure_natasha_en_AU_hd	Natasha (Australian)	en_AU	en-AU-NatashaNeural
azure_william_en_AU_hd	William (Australian)	en_AU	en-AU-WilliamNeural
azure_neerja_en_IN_hd	Neerja (Indian)	en_IN	en-IN-NeerjaNeural
azure_prabhat_en_IN_hd	Prabhat (Indian)	en_IN	en-IN-PrabhatNeural
azure_clara_en_CA_hd	Clara (Canadian)	en_CA	en-CA-ClaraNeural

Supertonic 3 (10 voices)

Newest local family (#1191, live v1.2.3). Ten en_US speakers from a shared ~139 MB model bundle (7 files, hosted on the VoxSherpa-TTS supertonic-v1 release) — downloaded once, like Kokoro. All High quality.

ID	Name	Gender	Speaker
supertonic_f1_en_US_0	Supertonic F1	F	0
supertonic_f2_en_US_1	Supertonic F2	F	1
supertonic_f3_en_US_2	Supertonic F3	F	2
supertonic_f4_en_US_3	Supertonic F4	F	3
supertonic_f5_en_US_4	Supertonic F5	F	4
supertonic_m1_en_US_5	Supertonic M1	M	5
supertonic_m2_en_US_6	Supertonic M2	M	6
supertonic_m3_en_US_7	Supertonic M3	M	7
supertonic_m4_en_US_8	Supertonic M4	M	8
supertonic_m5_en_US_9	Supertonic M5	M	9

Maintainer notes

• Source of truth: core-playback/src/main/kotlin/in/jphe/storyvox/playback/voice/VoiceCatalog.kt. Editing the catalog is a deliberate human review — scripts/voices/refresh-voices-v2.sh updates the hosted assets but does not edit VoiceCatalog.kt.
• The _int8 suffix in some Piper IDs is historical (kept stable across the v1 → v2 migration) — files served are fp32. See Voices page for the rationale.
• New upstream voices are detected monthly by .github/workflows/voice-catalog-check.yml and surfaced as a GitHub issue. Adding them is a manual edit + ship.
• Catalog drift between this wiki page and VoiceCatalog.kt is possible — the source file always wins. If counts here disagree, regenerate this page from the file.