Echo voices and Baobab national models — Africa-first language intelligence.

Explore Baobab
Skip to content

National model card

LugemiBaobab Ga

Structured status — not a raw JSON dump.

National Ga (Accra) speech + phrase MT corridor — generative_inference_live false; neural_nmt false; trainingEligible locked. ISO 639-3 gaa (not Irish ga).

Overview

Identity & capabilities

Model

national_shipped
model_id
baobab-ga
display_name
Baobab Ga
model_version
2026.10.national
family
baobab
status
national_shipped
as_of
2026-10-22T16:48:09.288Z
product_line
National Ga (Accra) speech + phrase MT corridor — generative_inference_live false; neural_nmt false; trainingEligible locked. ISO 639-3 gaa (not Irish ga).

capabilities

phrase_mt
phrase_table_only
national_api_tiers
true
generative_lm_research_slot
true
generative_inference_live
false
neural_nmt
false
neural_asr
false
neural_tts
false
s2s_corridor
true
foundation_llm_weights
false

honesty flags

Flags are published as returned by the API — unmet gates stay false.

trainingEligiblefalse
neural_nmtfalse
neural_speech_livefalse
generative_inference_livefalse
foundation_llm_weightsfalse
ships_trained_baobab_foundationfalse
phrase_table_onlytrue
irish_ga_interchangeablefalse
dangme_interchangeablefalse
variety_exclusive_speech_finetunesfalse
iso_code
gaa
bare_ga_warning
Irish `ga` / `ga-IE` is not Gã. Pass ISO 639-3 `gaa` or `gaa-GH` / variety `gaa-gh`. Dangme (`ada`) is never interchangeable.
license_warnings
  • Ga corpus TTS (ghanaopenai/ga-multispeaker-speech-text-20k unit bank) is CC-BY-NC-4.0 — unrestricted commercial redistribution of those assets is blocked.
  • Whisper-medium Ga ASR (obour/whisper-medium-ga-multispeaker-ghana) — pin versions and attribute per model card.
  • Meta MMS has Ga NLP only — facebook/mms-tts-gaa does not exist; do not invent MMS-TTS claims.
  • Phrase/lexical Baobab MT covers reviewed corridor domains; neural_nmt is false (no Ga NLLB service shipped).

varieties

Accra Ga variety

Accra / Ga Mashie only (ISO gaa). Not Irish ga, not Dangme.

accra

Ga (Accra)

gaa-gh
bcp47
gaa-GH
corridor_id
en-gaa-accra
primary_use
Greater Accra citizen services / education / MeetingBridge Ghana
language_pack_version
gaa-accra-pack-2026.10
echo_voice_id
own:gaa-gh-female
english_accent
own:en-gh-female · en-GH · Ama · Ghanaian English
mt_lang_key
gaa
trainingEligible
false
domains
  • greetings
  • everyday
  • airtime
  • health
  • civic
  • market
not_interchangeable_with
  • ada
  • ga
  • ga-IE
  • ak-gh-asante
  • ak-gh-akuapem

mt

status
phrase_table_only
pairs
en↔gaa

Deterministic phrase + lexical Baobab pack in lugemi-mt.engine. Fail-closed when uncovered. No dedicated Ga neural NMT service.

asr

status
neural_offline
checkpoint
obour/whisper-medium-ga-multispeaker-ghana
license
see_model_card
speech_stack
Ga speech
variety_exclusive_finetune
false

tts

status
neural_offline
checkpoint
ghanaopenai/ga-multispeaker-speech-text-20k unit bank
license
CC-BY-NC-4.0
speech_stack
Ga speech
variety_exclusive_finetune
false

s2s_corridor

status
corridor_pass

en-gaa-accra corridor scripts PASS when Ga speech is live. Broad domain deferred.

national_tiers

Who can call what

national_tiers

Public demo

public_demo
who
Citizens, press, students (unauthenticated demos)
auth
None — soft abuse controls on demo speech
maps_to_plan
null
audit
false
commercial_nc_clean_tts_required
false
endpoints
  • GET /v1/models/baobab-ga
  • GET /v1/baobab/ga/national
  • POST /v1/demo/speech
controls
  • honest en-GH remap when Ga speech is offline (paired Ghanaian English accent)
  • ISO gaa headers — never Irish ga
limits
translate_chars_daily_hint
Authenticated translate required for production MT
demo_speech
400 requests / 10 minutes / IP (demo controller)
rate_limit_note
Demo speech returns rate_limited when exceeded

national_tiers

Education / free

education
who
Students, UG / KNUST labs, research orgs on Free/Pro plans
auth
Org API key (TranslateAuthGuard) — Free or Pro plan
maps_to_plan
free|pro
audit
true
commercial_nc_clean_tts_required
false
residency_preference
GH
endpoints
  • POST /v1/translate
  • POST /v1/tts/synthesize
  • POST /v1/audio/transcriptions (when ASR keyed)
  • GET /v1/models/baobab-ga
controls
  • UsageService metering
  • BillingService.assertWithinQuota
limits
translate_chars_daily_hint
Free: BILLING_FREE_TRANSLATE_CHARS_QUOTA (default 50k); Pro: BILLING_PRO_TRANSLATE_CHARS_QUOTA
rate_limit_note
RATE_LIMIT_FREE_PER_KEY / RATE_LIMIT_PRO_PER_KEY via existing RateLimitGuard

national_tiers

Ministry production

ministry
who
Ghana government services (Ga / Greater Accra citizen channels)
auth
Ministry-keyed org API key — Business plan mapping
maps_to_plan
business
audit
true
commercial_nc_clean_tts_required
false
residency_preference
GH
endpoints
  • POST /v1/translate
  • POST /v1/tts/synthesize
  • MeetingBridge Ghana corridor
  • GET /v1/residency
  • GET /v1/metrics/translate
controls
  • AuditService on translate.completed
  • Org retention / vendor-training settings
  • Residency preference GH (affinity, not hard geo-fence)
limits
translate_chars_daily_hint
Business translate quota (BILLING_BUSINESS_TRANSLATE_CHARS_QUOTA)
rate_limit_note
RATE_LIMIT_BUSINESS_PER_KEY / PER_ORG

national_tiers

Enterprise

enterprise
who
Banks, telcos, NGOs serving Greater Accra
auth
Paid Enterprise keys
maps_to_plan
enterprise
audit
true
commercial_nc_clean_tts_required
true
residency_preference
AF
endpoints
  • POST /v1/translate (+ TM / quality)
  • Speech + MeetingBridge
  • Fine-tunes / marketplace where plan allows
controls
  • SLA targets (documented; contractual SLA separate)
  • Commercial TTS path must be NC-clean for redistribution of corpus TTS assets
  • Audit + retention
limits
translate_chars_daily_hint
Enterprise translate quota
rate_limit_note
RATE_LIMIT_ENTERPRISE_PER_KEY / PER_ORG

research · acceptance_gates

Generative LM research slot

Research slot + harness + training recipe only. Do not brand phrase MT or speech corridors as a full national LLM. Generative inference unsupported until gates are met and Lugemi-owned weights exist.

baobab-ga-generative-lm

research_planned
generative_inference_livefalse
foundation_llm_weightsfalse
ships_trained_baobab_foundationfalse
separate_variety_checkpoints_requiredtrue
proposed_checkpoint_ids.accra
baobab-ga-accra-lm-*
proposed_checkpoint_ids.never_conflate_dangme
true
proposed_checkpoint_ids.never_irish_ga
true

acceptance_gates

  • held_out_speakersrequired truemet false

    Held-out evaluation speakers must never appear in training exports; speaker lists remain empty until consented collection.

  • community_consent_accra_garequired truemet false

    trainingEligible stays false until Accra / Ga Mashie community speakers re-confirm for training.

  • dangme_separationrequired truemet false

    Any future LM/adapters must ship as baobab-ga-accra-* — never one blob marketed as Ga+Dangme.

  • beat_phrase_mt_baselinesrequired truemet false

    Generative claims require held-out gains vs corridor phrase-MT baselines on declared Ga-native tasks.

  • commercial_rights_clearrequired truemet false

    Base model + data licenses must clear commercial ministry/enterprise redistribution before promotion.

training_eligible_unlock

trainingEligiblefalse
statuslocked

trainingEligible remains false until all gates are met with durable evidence. Do not invent consent. Irish ga is never Ga.

  • community_consent_accra_garequired truemet false

    Accra / Ga Mashie community speakers must reconfirm training export consent.

    evidence_required Signed Accra Ga contributor agreements reconfirming training use

  • held_out_speaker_listsrequired truemet false

    Held-out evaluation speakers must never appear in training exports.

    evidence_required Published held-out speaker IDs never present in train splits

  • accra_ga_native_reviewrequired truemet false

    Native review must cover Accra Ga before any train export.

    evidence_required Paid Accra Ga native review records with trainingEligible fields

  • dangme_not_conflatedrequired truemet false

    Dangme is a separate language — never conflated with Gã training data.

    evidence_required Corpus metadata proving Dangme (ada) excluded from Ga train exports

  • rights_clear_for_exportrequired truemet false

    IP / commercial rights must clear before trainingEligible flips.

    evidence_required Legal memo clearing commercial vs research split for export corpus

honesty.residency

Serving & Accra affinity

serving_today
jnb
serving_today_label
Africa-primary Fly region Johannesburg (jnb)
affinity_target
af-west-1
affinity_city
Accra · GH
accra_gpu_island
false
accra_gpu_island_status
deferred
hard_geo_fence
false
preference_model
affinity_not_hard_fence
affinity_enforcement
soft_preference_recorded
note
Inference prefers matching person/org residency to nearby model hosting when ready — affinity, not a hard geo-fence. Accra (af-west-1) is the Ghana affinity target; live serving for verbalab + lugemi-ga is jnb today. No Accra GPU island is deployed. Ministry requests record X-Lugemi-Residency-Affinity soft pin.

endpoints · docs

API surface & references

endpoints

model_card
GET /v1/models/baobab-ga
national
GET /v1/baobab/ga/national
research
GET /v1/baobab/ga/research
unlock
GET /v1/baobab/ga/unlock
slo
GET /v1/baobab/ga/slo
translate
POST /v1/translate
demo_speech
POST /v1/demo/speech
tts
POST /v1/tts/synthesize
meetingbridge_ghana
MeetingBridge Ghana corridor
corridor
en-gaa-accra
residency
GET /v1/residency
translate_metrics
GET /v1/metrics/translate

Machine-readable card: https://api.lugemi.com/v1/models/baobab-ga

docs

national_model
docs/baobab/GA_NATIONAL_MODEL.md
model_card
docs/baobab/baobab-ga.MODEL_CARD.md
partnership_open_eval
docs/baobab/PARTNERSHIP_OPEN_EVAL.md
ministry_sla
docs/baobab/MINISTRY_SLA.md
training_eligible_unlock
docs/baobab/TRAINING_ELIGIBLE_UNLOCK.md
speech_finetune_recipe
docs/baobab/SPEECH_FINETUNE_RECIPE.md
generative_lm_recipe
docs/baobab/GENERATIVE_LM_RECIPE.md
accra_residency
docs/baobab/ACCRA_RESIDENCY.md
corridor
en-gaa-accra
language_kit
docs/models/05_LANGUAGE_KIT.md
residency
docs/residency.md

Try the corridor.

Run Translate, open the Baobab canopy, or fetch the same fields as JSON for clients.