mirror of
https://github.com/memohai/Memoh.git
synced 2026-04-25 07:00:48 +09:00
Feat/speech support (#392)
* feat: expand speech provider support with new client types and configuration schema * feat: add icon support for speech providers and update related configurations * feat: add SVG support for Deepgram and Elevenlabs with Vue components * feat: except *-speech client type in llm provider * feat: enhance speech provider functionality with advanced settings and model import capabilities * chore: remove go.mod replace * feat: enhance speech provider functionality with advanced settings and model import capabilities * chore: update go module dependencies * feat: Ear and Mouth * fix: separate ear/mouth page * fix: separate audio domain and restore transcription templates Move speech and transcription internals into the audio domain, restore template-driven transcription providers, and regenerate Swagger/SDK so the frontend can stop hand-calling /transcription-* APIs. --------- Co-authored-by: aki <arisu@ieee.org>
This commit is contained in:
Acbox
@@ -511,7 +511,7 @@ WITH updated AS (
|
||||
SET display_name = $1,
|
||||
updated_at = now()
|
||||
WHERE bots.id = $2
|
||||
RETURNING id, owner_user_id, display_name, avatar_url, timezone, is_active, status, language, reasoning_enabled, reasoning_effort, chat_model_id, search_provider_id, memory_provider_id, heartbeat_enabled, heartbeat_interval, heartbeat_prompt, heartbeat_model_id, compaction_enabled, compaction_threshold, compaction_ratio, compaction_model_id, title_model_id, image_model_id, discuss_probe_model_id, tts_model_id, browser_context_id, persist_full_tool_results, metadata, created_at, updated_at, acl_default_effect
|
||||
RETURNING id, owner_user_id, display_name, avatar_url, timezone, is_active, status, language, reasoning_enabled, reasoning_effort, chat_model_id, search_provider_id, memory_provider_id, heartbeat_enabled, heartbeat_interval, heartbeat_prompt, heartbeat_model_id, compaction_enabled, compaction_threshold, compaction_ratio, compaction_model_id, title_model_id, image_model_id, discuss_probe_model_id, tts_model_id, transcription_model_id, browser_context_id, persist_full_tool_results, metadata, created_at, updated_at, acl_default_effect
|
||||
)
|
||||
SELECT
|
||||
updated.id AS id,
|
||||
|
||||
@@ -34,6 +34,7 @@ type Bot struct {
|
||||
ImageModelID pgtype.UUID `json:"image_model_id"`
|
||||
DiscussProbeModelID pgtype.UUID `json:"discuss_probe_model_id"`
|
||||
TtsModelID pgtype.UUID `json:"tts_model_id"`
|
||||
TranscriptionModelID pgtype.UUID `json:"transcription_model_id"`
|
||||
BrowserContextID pgtype.UUID `json:"browser_context_id"`
|
||||
PersistFullToolResults bool `json:"persist_full_tool_results"`
|
||||
Metadata []byte `json:"metadata"`
|
||||
|
||||
+225
-11
@@ -13,7 +13,7 @@ import (
|
||||
|
||||
const countModels = `-- name: CountModels :one
|
||||
SELECT COUNT(*) FROM models
|
||||
WHERE type != 'speech'
|
||||
WHERE type NOT IN ('speech', 'transcription')
|
||||
`
|
||||
|
||||
func (q *Queries) CountModels(ctx context.Context) (int64, error) {
|
||||
@@ -40,13 +40,19 @@ FROM providers
|
||||
WHERE client_type NOT IN (
|
||||
'edge-speech',
|
||||
'openai-speech',
|
||||
'openai-transcription',
|
||||
'openrouter-speech',
|
||||
'openrouter-transcription',
|
||||
'elevenlabs-speech',
|
||||
'elevenlabs-transcription',
|
||||
'deepgram-speech',
|
||||
'deepgram-transcription',
|
||||
'minimax-speech',
|
||||
'volcengine-speech',
|
||||
'alibabacloud-speech',
|
||||
'microsoft-speech'
|
||||
'microsoft-speech',
|
||||
'google-speech',
|
||||
'google-transcription'
|
||||
)
|
||||
`
|
||||
|
||||
@@ -201,6 +207,24 @@ func (q *Queries) DeleteModelByModelID(ctx context.Context, modelID string) erro
|
||||
return err
|
||||
}
|
||||
|
||||
const deleteModelByProviderAndType = `-- name: DeleteModelByProviderAndType :exec
|
||||
DELETE FROM models
|
||||
WHERE provider_id = $1
|
||||
AND model_id = $2
|
||||
AND type = $3
|
||||
`
|
||||
|
||||
type DeleteModelByProviderAndTypeParams struct {
|
||||
ProviderID pgtype.UUID `json:"provider_id"`
|
||||
ModelID string `json:"model_id"`
|
||||
Type string `json:"type"`
|
||||
}
|
||||
|
||||
func (q *Queries) DeleteModelByProviderAndType(ctx context.Context, arg DeleteModelByProviderAndTypeParams) error {
|
||||
_, err := q.db.Exec(ctx, deleteModelByProviderAndType, arg.ProviderID, arg.ModelID, arg.Type)
|
||||
return err
|
||||
}
|
||||
|
||||
const deleteModelByProviderIDAndModelID = `-- name: DeleteModelByProviderIDAndModelID :exec
|
||||
DELETE FROM models
|
||||
WHERE provider_id = $1
|
||||
@@ -294,6 +318,27 @@ func (q *Queries) GetModelByProviderAndModelID(ctx context.Context, arg GetModel
|
||||
return i, err
|
||||
}
|
||||
|
||||
const getProviderByClientType = `-- name: GetProviderByClientType :one
|
||||
SELECT id, name, client_type, icon, enable, config, metadata, created_at, updated_at FROM providers WHERE client_type = $1
|
||||
`
|
||||
|
||||
func (q *Queries) GetProviderByClientType(ctx context.Context, clientType string) (Provider, error) {
|
||||
row := q.db.QueryRow(ctx, getProviderByClientType, clientType)
|
||||
var i Provider
|
||||
err := row.Scan(
|
||||
&i.ID,
|
||||
&i.Name,
|
||||
&i.ClientType,
|
||||
&i.Icon,
|
||||
&i.Enable,
|
||||
&i.Config,
|
||||
&i.Metadata,
|
||||
&i.CreatedAt,
|
||||
&i.UpdatedAt,
|
||||
)
|
||||
return i, err
|
||||
}
|
||||
|
||||
const getProviderByID = `-- name: GetProviderByID :one
|
||||
SELECT id, name, client_type, icon, enable, config, metadata, created_at, updated_at FROM providers WHERE id = $1
|
||||
`
|
||||
@@ -375,12 +420,51 @@ func (q *Queries) GetSpeechModelWithProvider(ctx context.Context, id pgtype.UUID
|
||||
return i, err
|
||||
}
|
||||
|
||||
const getTranscriptionModelWithProvider = `-- name: GetTranscriptionModelWithProvider :one
|
||||
SELECT
|
||||
m.id, m.model_id, m.name, m.provider_id, m.type, m.config, m.created_at, m.updated_at,
|
||||
p.client_type AS provider_type
|
||||
FROM models m
|
||||
JOIN providers p ON p.id = m.provider_id
|
||||
WHERE m.id = $1
|
||||
AND m.type = 'transcription'
|
||||
`
|
||||
|
||||
type GetTranscriptionModelWithProviderRow struct {
|
||||
ID pgtype.UUID `json:"id"`
|
||||
ModelID string `json:"model_id"`
|
||||
Name pgtype.Text `json:"name"`
|
||||
ProviderID pgtype.UUID `json:"provider_id"`
|
||||
Type string `json:"type"`
|
||||
Config []byte `json:"config"`
|
||||
CreatedAt pgtype.Timestamptz `json:"created_at"`
|
||||
UpdatedAt pgtype.Timestamptz `json:"updated_at"`
|
||||
ProviderType string `json:"provider_type"`
|
||||
}
|
||||
|
||||
func (q *Queries) GetTranscriptionModelWithProvider(ctx context.Context, id pgtype.UUID) (GetTranscriptionModelWithProviderRow, error) {
|
||||
row := q.db.QueryRow(ctx, getTranscriptionModelWithProvider, id)
|
||||
var i GetTranscriptionModelWithProviderRow
|
||||
err := row.Scan(
|
||||
&i.ID,
|
||||
&i.ModelID,
|
||||
&i.Name,
|
||||
&i.ProviderID,
|
||||
&i.Type,
|
||||
&i.Config,
|
||||
&i.CreatedAt,
|
||||
&i.UpdatedAt,
|
||||
&i.ProviderType,
|
||||
)
|
||||
return i, err
|
||||
}
|
||||
|
||||
const listEnabledModels = `-- name: ListEnabledModels :many
|
||||
SELECT m.id, m.model_id, m.name, m.provider_id, m.type, m.config, m.created_at, m.updated_at
|
||||
FROM models m
|
||||
JOIN providers p ON m.provider_id = p.id
|
||||
WHERE p.enable = true
|
||||
AND m.type != 'speech'
|
||||
AND m.type NOT IN ('speech', 'transcription')
|
||||
ORDER BY m.created_at DESC
|
||||
`
|
||||
|
||||
@@ -525,7 +609,7 @@ func (q *Queries) ListModelVariantsByModelUUID(ctx context.Context, modelUuid pg
|
||||
|
||||
const listModels = `-- name: ListModels :many
|
||||
SELECT id, model_id, name, provider_id, type, config, created_at, updated_at FROM models
|
||||
WHERE type != 'speech'
|
||||
WHERE type NOT IN ('speech', 'transcription')
|
||||
ORDER BY created_at DESC
|
||||
`
|
||||
|
||||
@@ -633,7 +717,7 @@ func (q *Queries) ListModelsByProviderClientType(ctx context.Context, clientType
|
||||
const listModelsByProviderID = `-- name: ListModelsByProviderID :many
|
||||
SELECT id, model_id, name, provider_id, type, config, created_at, updated_at FROM models
|
||||
WHERE provider_id = $1
|
||||
AND type != 'speech'
|
||||
AND type NOT IN ('speech', 'transcription')
|
||||
ORDER BY created_at DESC
|
||||
`
|
||||
|
||||
@@ -747,13 +831,19 @@ SELECT id, name, client_type, icon, enable, config, metadata, created_at, update
|
||||
WHERE client_type NOT IN (
|
||||
'edge-speech',
|
||||
'openai-speech',
|
||||
'openai-transcription',
|
||||
'openrouter-speech',
|
||||
'openrouter-transcription',
|
||||
'elevenlabs-speech',
|
||||
'elevenlabs-transcription',
|
||||
'deepgram-speech',
|
||||
'deepgram-transcription',
|
||||
'minimax-speech',
|
||||
'volcengine-speech',
|
||||
'alibabacloud-speech',
|
||||
'microsoft-speech'
|
||||
'microsoft-speech',
|
||||
'google-speech',
|
||||
'google-transcription'
|
||||
)
|
||||
ORDER BY created_at DESC
|
||||
`
|
||||
@@ -921,6 +1011,135 @@ func (q *Queries) ListSpeechProviders(ctx context.Context) ([]Provider, error) {
|
||||
return items, nil
|
||||
}
|
||||
|
||||
const listTranscriptionModels = `-- name: ListTranscriptionModels :many
|
||||
SELECT m.id, m.model_id, m.name, m.provider_id, m.type, m.config, m.created_at, m.updated_at,
|
||||
p.client_type AS provider_type
|
||||
FROM models m
|
||||
JOIN providers p ON p.id = m.provider_id
|
||||
WHERE m.type = 'transcription'
|
||||
ORDER BY m.created_at DESC
|
||||
`
|
||||
|
||||
type ListTranscriptionModelsRow struct {
|
||||
ID pgtype.UUID `json:"id"`
|
||||
ModelID string `json:"model_id"`
|
||||
Name pgtype.Text `json:"name"`
|
||||
ProviderID pgtype.UUID `json:"provider_id"`
|
||||
Type string `json:"type"`
|
||||
Config []byte `json:"config"`
|
||||
CreatedAt pgtype.Timestamptz `json:"created_at"`
|
||||
UpdatedAt pgtype.Timestamptz `json:"updated_at"`
|
||||
ProviderType string `json:"provider_type"`
|
||||
}
|
||||
|
||||
func (q *Queries) ListTranscriptionModels(ctx context.Context) ([]ListTranscriptionModelsRow, error) {
|
||||
rows, err := q.db.Query(ctx, listTranscriptionModels)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
defer rows.Close()
|
||||
var items []ListTranscriptionModelsRow
|
||||
for rows.Next() {
|
||||
var i ListTranscriptionModelsRow
|
||||
if err := rows.Scan(
|
||||
&i.ID,
|
||||
&i.ModelID,
|
||||
&i.Name,
|
||||
&i.ProviderID,
|
||||
&i.Type,
|
||||
&i.Config,
|
||||
&i.CreatedAt,
|
||||
&i.UpdatedAt,
|
||||
&i.ProviderType,
|
||||
); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
items = append(items, i)
|
||||
}
|
||||
if err := rows.Err(); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
return items, nil
|
||||
}
|
||||
|
||||
const listTranscriptionModelsByProviderID = `-- name: ListTranscriptionModelsByProviderID :many
|
||||
SELECT id, model_id, name, provider_id, type, config, created_at, updated_at FROM models
|
||||
WHERE provider_id = $1
|
||||
AND type = 'transcription'
|
||||
ORDER BY created_at DESC
|
||||
`
|
||||
|
||||
func (q *Queries) ListTranscriptionModelsByProviderID(ctx context.Context, providerID pgtype.UUID) ([]Model, error) {
|
||||
rows, err := q.db.Query(ctx, listTranscriptionModelsByProviderID, providerID)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
defer rows.Close()
|
||||
var items []Model
|
||||
for rows.Next() {
|
||||
var i Model
|
||||
if err := rows.Scan(
|
||||
&i.ID,
|
||||
&i.ModelID,
|
||||
&i.Name,
|
||||
&i.ProviderID,
|
||||
&i.Type,
|
||||
&i.Config,
|
||||
&i.CreatedAt,
|
||||
&i.UpdatedAt,
|
||||
); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
items = append(items, i)
|
||||
}
|
||||
if err := rows.Err(); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
return items, nil
|
||||
}
|
||||
|
||||
const listTranscriptionProviders = `-- name: ListTranscriptionProviders :many
|
||||
SELECT id, name, client_type, icon, enable, config, metadata, created_at, updated_at FROM providers
|
||||
WHERE client_type IN (
|
||||
'openai-transcription',
|
||||
'openrouter-transcription',
|
||||
'elevenlabs-transcription',
|
||||
'deepgram-transcription',
|
||||
'google-transcription'
|
||||
)
|
||||
ORDER BY created_at DESC
|
||||
`
|
||||
|
||||
func (q *Queries) ListTranscriptionProviders(ctx context.Context) ([]Provider, error) {
|
||||
rows, err := q.db.Query(ctx, listTranscriptionProviders)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
defer rows.Close()
|
||||
var items []Provider
|
||||
for rows.Next() {
|
||||
var i Provider
|
||||
if err := rows.Scan(
|
||||
&i.ID,
|
||||
&i.Name,
|
||||
&i.ClientType,
|
||||
&i.Icon,
|
||||
&i.Enable,
|
||||
&i.Config,
|
||||
&i.Metadata,
|
||||
&i.CreatedAt,
|
||||
&i.UpdatedAt,
|
||||
); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
items = append(items, i)
|
||||
}
|
||||
if err := rows.Err(); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
return items, nil
|
||||
}
|
||||
|
||||
const updateModel = `-- name: UpdateModel :one
|
||||
UPDATE models
|
||||
SET
|
||||
@@ -1062,11 +1281,6 @@ VALUES ($1, $2, $3, false, $4, '{}')
|
||||
ON CONFLICT (name) DO UPDATE SET
|
||||
icon = EXCLUDED.icon,
|
||||
client_type = EXCLUDED.client_type,
|
||||
config = CASE
|
||||
WHEN providers.config->>'api_key' IS NOT NULL AND providers.config->>'api_key' != ''
|
||||
THEN jsonb_set(EXCLUDED.config, '{api_key}', providers.config->'api_key')
|
||||
ELSE EXCLUDED.config
|
||||
END,
|
||||
updated_at = now()
|
||||
RETURNING id, name, client_type, icon, enable, config, metadata, created_at, updated_at
|
||||
`
|
||||
|
||||
@@ -30,6 +30,7 @@ SET language = 'auto',
|
||||
search_provider_id = NULL,
|
||||
memory_provider_id = NULL,
|
||||
tts_model_id = NULL,
|
||||
transcription_model_id = NULL,
|
||||
browser_context_id = NULL,
|
||||
persist_full_tool_results = false,
|
||||
updated_at = now()
|
||||
@@ -62,6 +63,7 @@ SELECT
|
||||
memory_providers.id AS memory_provider_id,
|
||||
image_models.id AS image_model_id,
|
||||
tts_models.id AS tts_model_id,
|
||||
transcription_models.id AS transcription_model_id,
|
||||
browser_contexts.id AS browser_context_id,
|
||||
bots.persist_full_tool_results
|
||||
FROM bots
|
||||
@@ -73,6 +75,7 @@ LEFT JOIN models AS image_models ON image_models.id = bots.image_model_id
|
||||
LEFT JOIN search_providers ON search_providers.id = bots.search_provider_id
|
||||
LEFT JOIN memory_providers ON memory_providers.id = bots.memory_provider_id
|
||||
LEFT JOIN models AS tts_models ON tts_models.id = bots.tts_model_id
|
||||
LEFT JOIN models AS transcription_models ON transcription_models.id = bots.transcription_model_id
|
||||
LEFT JOIN browser_contexts ON browser_contexts.id = bots.browser_context_id
|
||||
WHERE bots.id = $1
|
||||
`
|
||||
@@ -97,6 +100,7 @@ type GetSettingsByBotIDRow struct {
|
||||
MemoryProviderID pgtype.UUID `json:"memory_provider_id"`
|
||||
ImageModelID pgtype.UUID `json:"image_model_id"`
|
||||
TtsModelID pgtype.UUID `json:"tts_model_id"`
|
||||
TranscriptionModelID pgtype.UUID `json:"transcription_model_id"`
|
||||
BrowserContextID pgtype.UUID `json:"browser_context_id"`
|
||||
PersistFullToolResults bool `json:"persist_full_tool_results"`
|
||||
}
|
||||
@@ -124,6 +128,7 @@ func (q *Queries) GetSettingsByBotID(ctx context.Context, id pgtype.UUID) (GetSe
|
||||
&i.MemoryProviderID,
|
||||
&i.ImageModelID,
|
||||
&i.TtsModelID,
|
||||
&i.TranscriptionModelID,
|
||||
&i.BrowserContextID,
|
||||
&i.PersistFullToolResults,
|
||||
)
|
||||
@@ -151,11 +156,12 @@ WITH updated AS (
|
||||
memory_provider_id = COALESCE($16::uuid, bots.memory_provider_id),
|
||||
image_model_id = COALESCE($17::uuid, bots.image_model_id),
|
||||
tts_model_id = COALESCE($18::uuid, bots.tts_model_id),
|
||||
browser_context_id = COALESCE($19::uuid, bots.browser_context_id),
|
||||
persist_full_tool_results = $20,
|
||||
transcription_model_id = COALESCE($19::uuid, bots.transcription_model_id),
|
||||
browser_context_id = COALESCE($20::uuid, bots.browser_context_id),
|
||||
persist_full_tool_results = $21,
|
||||
updated_at = now()
|
||||
WHERE bots.id = $21
|
||||
RETURNING bots.id, bots.language, bots.reasoning_enabled, bots.reasoning_effort, bots.heartbeat_enabled, bots.heartbeat_interval, bots.heartbeat_prompt, bots.compaction_enabled, bots.compaction_threshold, bots.compaction_ratio, bots.timezone, bots.chat_model_id, bots.heartbeat_model_id, bots.compaction_model_id, bots.title_model_id, bots.image_model_id, bots.search_provider_id, bots.memory_provider_id, bots.tts_model_id, bots.browser_context_id, bots.persist_full_tool_results
|
||||
WHERE bots.id = $22
|
||||
RETURNING bots.id, bots.language, bots.reasoning_enabled, bots.reasoning_effort, bots.heartbeat_enabled, bots.heartbeat_interval, bots.heartbeat_prompt, bots.compaction_enabled, bots.compaction_threshold, bots.compaction_ratio, bots.timezone, bots.chat_model_id, bots.heartbeat_model_id, bots.compaction_model_id, bots.title_model_id, bots.image_model_id, bots.search_provider_id, bots.memory_provider_id, bots.tts_model_id, bots.transcription_model_id, bots.browser_context_id, bots.persist_full_tool_results
|
||||
)
|
||||
SELECT
|
||||
updated.id AS bot_id,
|
||||
@@ -177,6 +183,7 @@ SELECT
|
||||
memory_providers.id AS memory_provider_id,
|
||||
image_models.id AS image_model_id,
|
||||
tts_models.id AS tts_model_id,
|
||||
transcription_models.id AS transcription_model_id,
|
||||
browser_contexts.id AS browser_context_id,
|
||||
updated.persist_full_tool_results
|
||||
FROM updated
|
||||
@@ -188,6 +195,7 @@ LEFT JOIN models AS image_models ON image_models.id = updated.image_model_id
|
||||
LEFT JOIN search_providers ON search_providers.id = updated.search_provider_id
|
||||
LEFT JOIN memory_providers ON memory_providers.id = updated.memory_provider_id
|
||||
LEFT JOIN models AS tts_models ON tts_models.id = updated.tts_model_id
|
||||
LEFT JOIN models AS transcription_models ON transcription_models.id = updated.transcription_model_id
|
||||
LEFT JOIN browser_contexts ON browser_contexts.id = updated.browser_context_id
|
||||
`
|
||||
|
||||
@@ -210,6 +218,7 @@ type UpsertBotSettingsParams struct {
|
||||
MemoryProviderID pgtype.UUID `json:"memory_provider_id"`
|
||||
ImageModelID pgtype.UUID `json:"image_model_id"`
|
||||
TtsModelID pgtype.UUID `json:"tts_model_id"`
|
||||
TranscriptionModelID pgtype.UUID `json:"transcription_model_id"`
|
||||
BrowserContextID pgtype.UUID `json:"browser_context_id"`
|
||||
PersistFullToolResults bool `json:"persist_full_tool_results"`
|
||||
ID pgtype.UUID `json:"id"`
|
||||
@@ -235,6 +244,7 @@ type UpsertBotSettingsRow struct {
|
||||
MemoryProviderID pgtype.UUID `json:"memory_provider_id"`
|
||||
ImageModelID pgtype.UUID `json:"image_model_id"`
|
||||
TtsModelID pgtype.UUID `json:"tts_model_id"`
|
||||
TranscriptionModelID pgtype.UUID `json:"transcription_model_id"`
|
||||
BrowserContextID pgtype.UUID `json:"browser_context_id"`
|
||||
PersistFullToolResults bool `json:"persist_full_tool_results"`
|
||||
}
|
||||
@@ -259,6 +269,7 @@ func (q *Queries) UpsertBotSettings(ctx context.Context, arg UpsertBotSettingsPa
|
||||
arg.MemoryProviderID,
|
||||
arg.ImageModelID,
|
||||
arg.TtsModelID,
|
||||
arg.TranscriptionModelID,
|
||||
arg.BrowserContextID,
|
||||
arg.PersistFullToolResults,
|
||||
arg.ID,
|
||||
@@ -284,6 +295,7 @@ func (q *Queries) UpsertBotSettings(ctx context.Context, arg UpsertBotSettingsPa
|
||||
&i.MemoryProviderID,
|
||||
&i.ImageModelID,
|
||||
&i.TtsModelID,
|
||||
&i.TranscriptionModelID,
|
||||
&i.BrowserContextID,
|
||||
&i.PersistFullToolResults,
|
||||
)
|
||||
|
||||
Reference in New Issue
Block a user